US 20050267935 A1
A general programmatic interface-to-network messaging adapter exposes a suitable object integration interface or application programming interface to applications on a controller device and sends network data messages to invoke services or query status of a controlled device. The adapter maps application calls to the interface into network data messages according to service protocols of the controlled device. The general adapter provides the interface suitable to any specific service of a controlled device based on a data description of the interface, and converts the application calls to network data messages based on a data description of a protocol and format for network data messages to interact with the specific service. Once the interface/messaging description is obtained, applications on the controller device can programmatically interact with the adapter, and the adapter then handles appropriate message exchanges with the service of the controlled device. The general adapter allows controller device applications to be written using object-oriented programming, while avoiding code download.
3. A computing device providing a user control point with connectivity to at least one controlled device via a networking medium, the computing device comprising:
memory having stored thereon a controlled device description artifact having a service control definition and a contract for configuring communications involved with at least one service provided by the at least one controlled device, wherein the contract comprises one or more of a definition of network packets, request-response patterns and specification of an endpoint to which the communications comprising networking messages are sent;
a general programming interface-to-network messaging adapter operable as follows:
based on the service control definition in the controlled device description artifact, to dynamically produce a service specific programming interface to application programs running on the computing device,
to convert calls from the application programs to the service specific programming interface into the networking messages to the controlled device according to the contract specified in the controlled device description artifact; and
to issue the networking messages via the networking medium to the controlled device to invoke service specific operations of the at least one service.
4. The device of
5. The device of
6. The device of
7. The device of
8. The device of
9. The device of
10. A method for a client program on a first computing device to programmatically control a service of a logical device realized on a remote computing device on a data communications network via peer-to-peer networking connectivity from the first computing device on the data communications network, the method comprising:
obtaining a device description module having a textual description of the logical device service via peer-to-peer networking from the remote computing device, the device description module defining the logical device service in text and a contract also in text for an exchange of data messages via peer-to-peer networking connectivity with the remote computing device for controlling the logical device service on the remote computing device;
based on the textual description of the logical device service in the device description module, dynamically generating code of a programmatic interface for invocation by the client program to initiate service-specific operations for remote control of the logical device service; and
on invocation of method members of the generated programmatic interface by the client program, converting the client program's programmatic interface invocation into the exchange of data messages via peer-to-peer networking connectivity in accordance with the contract defined in the device description entity for effecting control of the logical device service.
11. The method of
12. The method of
13. The method of
14. The method of
15. The method of
16. The method of
17. A method for a client program on a first computing device to programmatically control a service of a logical device realized on a remote computing device over a data communications network via peer-to-peer networking connectivity from the first computing device on the data communications network, the method comprising:
obtaining device description data via peer-to-peer networking from the remote computing device, the device description data comprising a service-specific service control definition and a service-specific contract for an exchange of data messages via peer-to-peer networking connectivity with the remote computing device for controlling the logical device service on the remote computing device;
based on the service-specific service control definition in the device description data, dynamically producing an instance of a programmatic interface for invocation by the client program to initiate service-specific operations for remote control of the logical device service, the programmatic interface comprising an invoke method member operable for accepting a string argument specifying a service action associated with the logical device service; and
upon a calling of the invoke method member of the programmatic interface by the client program, translating the invoke method call from the client program into the exchange of data messages via peer-to-peer networking connectivity in accordance with the contract defined in the description document for effecting control of the logical device service.
18. The method of
19. The method of
20. The method of
21. The method of
22. The method of
This application is based on provisional application No. 60/139,137 filed and Jun. 11, 1999, and provisional application number [to be determined] filed Oct. 18, 1999.
This invention relates generally to dynamic configuration of interconnectivity among distributed devices and services, and more particularly relates to providing a capability to access device- or service-specific operational information and perform remote automation and control of embedded computing devices using a data-driven remote programming model, such as in a pervasive computing environment
The cost of computing and networking technologies have fallen to the point where computing and networking capabilities can be built into the design of many electronic devices in the home, the office and public places. The combination of inexpensive and reliable shared networking media with a new class of small computing devices has created an opportunity for new functionality based mainly on the connectivity among these devices. This connectivity can be used to remotely control devices, to move digital data in the form of audio, video and still images between devices, to share information among devices and with the unconstrained World Wide Web of the Internet (hereafter “Web”) and to exchange structured and secure digital data to support things like electronic commerce. The connectivity also enables many new applications for computing devices, such as proximity-based usage scenarios where devices interact based at least in part on geographical or other notions of proximity. A prevalent feature of these connectivity scenarios is to provide remote access and control of connected devices and services from another device with user interface capabilities (e.g., a universal remote controller, handheld computer or digital assistant, cell phones, and the like). These developments are occurring at the same time as more people are becoming connected to the Internet and as connectivity solutions are falling in price and increasing in speed. These trends are leading towards a world of ubiquitous and pervasive networked computing, where all types of devices are able to effortlessly and seamlessly interconnect and interact.
In accordance with a new device connectivity architecture known as Universal Plug and Play, devices and services are controlled by exchanging well-defined XML-format data messages. At the programmatic level, on the other hand, it is useful and productive to work in an object-oriented framework.
Prior connectivity models are not adequate to bridge between object interfaces and the data messages exchanged with the controlled device over a network. Some prior connectivity models require a controlling device to download the program code (such as a device driver, Jini code, etc.) for interacting with the controlled device or service from a networked source. Such a code download requirement is unsuitable to the Web and other ubiquitous computing scenarios. Other connectivity models require use of a custom-written object for specific classes of services. This approach leads to deployment hassles (e.g., user setup and configuration) and also is unsuitable to ubiquitous computing.
In accordance with a technology described herein, a general programmatic interface-to-network messaging adapter (called a “rehydrator”) is a module that exposes a suitable object integration interface or application programming interface to applications on a controller device and sends network data messages to invoke services or query status of a controlled device. The adapter maps application calls to the interface into network data messages according to service protocols of the controlled device. The described adapter preferably is generic to all devices and services compatible with the connectivity model, and adapts itself to specific of the devices based on an interface and message format/protocol description. In other words, this adapter operates as a universal module through which network data message-driven services on other networked computing devices can remote programmatic application programming interfaces, including object integration interfaces according to an object model such as Microsoft's COM, CORBA, JAVA, and the like.
More specifically, this general adapter provides the interface suitable to any specific service of a controlled device based on a data description of the interface, and converts the application calls to network data messages based on a data description of a protocol and format for network data messages to interact with the specific service. Once the interface/messaging description is obtained, applications on the controller device can programmatically interact with the adapter, and the adapter then handles appropriate message exchanges with the service of the controlled device. With the described adapter, no code download is required, only the interface/messaging description is needed. The description can be obtained from the controlled device, a network server computer, or by pre-loading or caching on the controller device. The technology allows controller device applications to be written using object-oriented programming, while avoiding code download.
Additional features and advantages will be made apparent from the following detailed description of the illustrated embodiment which proceeds with reference to the accompanying drawings.
The following detailed description is directed toward a general programmatic interface-to-network messaging adapter (also known as a “rehydrator”) in a device control model. In one described implementation, the rehydrator is used in a device architecture 100 (
Universal Plug and Play
Universal Plug and Play (UPnP) is an open network architecture that is designed to enable simple, ad hoc communication among distributed devices and services from many vendors. UPnP leverages Internet technology and can be thought of as an extension of the Web model of mobile web browsers talking to fixed web servers to the world of peer-to-peer connectivity among mobile and fixed devices. UPnP embraces the zero configuration mantra of Plug and Play (PnP) but is not a simple extension of the PnP host/peripheral model.
The cost, size and battery consumption of computing technology—including processing, storage and displays—continues to fall. This trend is enabling the evolution of stand-alone, single or limited function computing devices such as digital cameras, audio playback devices, smart mobile phones and handheld computers. Concurrent with this, the economical storage and transmission of digital audio, video and still images is enabling highly flexible models for managing entertainment content.
While many of these devices are capable of useful stand-alone operation, seamless connectivity with the PC can enhance the value to the customer of both stand-alone devices and the PC. Good examples of this synergy are digital image capture combined with PC image manipulation, storage and email transfer/web publishing and information synchronization between a PC and a handheld computer or smart mobile phone.
Since many of these devices, and the PC itself, are mobile, a suitable communication architecture must enable a highly dynamic connectivity model and must enable peer-to-peer operating among arbitrary combinations of devices.
The Internet has created a widespread awareness of the value of simple, universal communication that is independent of the underlying transmission technology and independent of technology from any single vendor.
UPnP makes it possible to initiate and control the transfer of bulk data (e.g. files) or A/V data streams from any device on the network, to any device on the network, under the control of any device on the network. UPnP enables the ad hoc addition or removal of devices on the network, and it enables multiple controlling devices to remain in sync with each other.
UPnP reuses existing protocols and technology whenever possible. The transition to this highly connected (and connectable) world will not occur overnight. UPnP builds on existing Internet protocols, but accommodates devices that cannot run the complete UPnP protocol suite. UPnP provides an architecture that enables legacy devices to communicate with UPnP devices.
IP internetworking has been chosen as a UPnP baseline due to its proven ability to span different physical media, to enable real world multiple vendor interoperation and to achieve synergy with the Internet and home and office intranets. Internet synergy enables applications such as IP telephony, multiple player games, remote control of home automation and security, Internet based electronic commerce, in addition to simple email and web browsing. UPnP's scope includes remote control of devices and bulk data transfer. But, it does not specify A/V streaming formats or protocols.
UPnP's media independence enables a great deal of flexibility in the packaging of products. UPnP enables an A/V system to be controlled through an A/C power communications technology, while the transmission of A/V streams among the components is analog or digital. One of the controllers of this system could be on the television, while another is on a PC, and yet another connected via radio or infrared.
Unlike Plug and Play, Universal Plug and Play is built on top of networking and enables ad hoc peer-to-peer connectivity. Networking, in this context, describes a style of connectivity that enables any networked device to initiate a communication with any other networked device, without having established a prior relationship or maintaining a persistent relationship between the devices. Networking also allows multiple devices to establish one or more connections with a single device, and it allows for a device to be capable of both initiating and accepting connections to/from other devices. The PnP, or host/peripheral, model is suitable whenever there is a natural persistent relationship between two devices (e.g. a keyboard, mouse and display maintain and a persistent relationship with a host computer). Even though networking does not mandate low level persistent relationships, it provides the needed anchors (addresses) for applications to choose to maintain associations as a convenience for the customer (e.g. remembering commonly used networked printers).
In order to achieve multiple vendor peer-to-peer interoperation among devices, vendors desirably agree on common technology and standards up to the highest level of desired functional interoperation.
UPnP leverages formal protocol contracts to enable peer-to-peer interoperation. Protocols contracts enable real-world multiple-vendor interoperation.
UPnP enables devices to expose a user interface by leveraging browser technology. In this context, the browser can be considered to be a very rich remote terminal. Current browser technology does not maintain a separation of presentation from data, or in the case of devices, control. It is possible to hunt through a page of HTML to extract data values, but it is not convenient or robust. UPnP leverages the separation of presentation and data enabled by the use of XML, and it extends this technology to the device control domain.
UPnP provides a device-driven auto-configuration capability that preserves the experience that customers have on the web. Today, it is possible to navigate around the web without loading programs beyond the browser itself. Since UPnP enables the browser to be extended to control devices, and because UPnP devices are controlled with explicit protocols, the browser must somehow learn how to talk to UPnP devices. This learning process is driven entirely from the device itself and is accomplishing entirely by uploading an XML document that describes the capabilities of the device. The architectural component that enables device-driven auto-configuration is called the Rehydrator. The job of the Rehydrator is to convert between APIs and protocols.
Since the auto-configuration process itself is driven only by the exchange of formatted data, there is very little opportunity for a malicious attack from a hostile piece of code.
There are some scenarios where the web UI model is not sufficient for a rich customer experience. It would not be convenient to have to a web for each light switch in a house. To support a rich user interface and to enable the aggregation of devices into a single UI, UPnP enables application control in addition to browser control of devices. This is achieved simply by enabling applications to call the same Rehydrator APIs that the browser does. Applications can also directly generate and consume the raw UPnP control protocols, provided they are not interested in the device-driven auto-configuration enabled by the Rehydrator.
UPnP assumes that there will be more than one device with UI that wants to control other devices in any given network, and it provides a simple mechanism that enables these control points to remain in sync. This mechanism can easily support device front panels and wireless remotes that do not run UPnP protocols. The UPnP control model is third-party control; any device can transfer bulk data (e.g. files) or A/V data streams from any device on the network, to any device on the network, under the control of any device on the network.
The detailed description that follows uses the terminology defined below.
Module. A component of a device, software program, or system that implements some “functionality”, which can be embodied as software, hardware, firmware, electronic circuitry, or etc.
User Control Point. The set of modules that enable communication with a UPnP Controlled Device. User Control Points initiate discovery and communication with Controlled Devices, and receive Events from Controlled Devices. User Control Points are typically implemented on devices that have a user interface. This user interface is used to interact with Controlled Devices over the network. The modules minimally include a Discovery Client, a Description Client and a Rehydrator. User Control Points may also include Visual Navigation, an Event Subscription Client, Event Sink, a web browser and an application execution environment. User Control Points can add value to the network by aggregating the control of multiple Controlled Devices (the universal remote) or they can implement a function as simple as initiating the transfer of data to or from a Controlled Device. Examples of devices that could be User Control Points are the personal computer (PC), digital television (DTV), set-top box (STB), handheld computer and smart mobile phone, and the like. Nothing prevents a single device from implementing the functionality of a User Control Point and one or more Controlled Devices at the same time.
Controlled Device. The set of modules that enable communication with a User Control Point. Controlled Devices respond to discovery requests, accept incoming communications from User Control Points and may send Events to User Control Points. Devices that support Controlled Device functionality may also support local user interfaces such as front panel displays or wireless remotes. The modules minimally include a Discovery Server, a Description Server and a Control Server. Controlled Devices may also include a Presentation (web) Server, Event Subscription Server and Event Source. Examples of devices that could be Controlled Devices are the VCR, DVD player or recorder, heating/ventilation/air-conditioning equipment (HVAC), lighting controller, audio/video/imaging playback device, handheld computer, smart mobile phone and the PC, and the like. Nothing prevents a single device from implementing the functionality of a User Control Point and one or more Controlled Devices at the same time.
Bridge. A set of modules that enables Bridged and Legacy Devices to interact with native UPnP devices. The bridge itself exposes a collection of UPnP Controlled Devices to User Control Points. The Bridge maps between native UPnP Device Control Protocols and the underlying protocols exposed by the Bridged and Legacy Devices. Optionally, such a device could expose UPnP Controlled Devices to Legacy Devices in the manner required by the Legacy Devices. Nothing prevents a single device from implementing the functionality of a User Control Point, one or more Controlled Devices and a Bridge at the same time.
Service Provider. A module used by a UPnP Bridge that translates between UPnP protocols and the protocols used by Bridged and Legacy Devices. No Service Providers are required for communication among native UPnP devices.
Bridged Device. A device that cannot participate in UPnP at the native protocol level, either because the device does not have sufficient resources or because the underlying media is unsuitable to run TCP and HTTP. Examples of devices that could be Bridged Devices are power line-controlled A/V equipment, light switches, thermostats, wristwatches and inexpensive toys. Bridged Devices are UPnP complaint and are exposed to other UPnP devices through a UPnP Bridge.
Legacy Device. Any non-UPnP compliant device that must be exposed to other UPnP devices through a UPnP Bridge.
Device Model. The UPnP model of Controlled Devices. The Device Model includes the addressing scheme, Description Document, Devices and Services hierarchy and the functional description of modules.
Device Control Protocol (DCP). A complete set of UPnP protocols and schemas used to interact with a UPnP Controlled Device.
Device Definition. The formal definition of a Device Type. A Device Definition includes a Device Type Identifier, the fixed elements in the Description Document, the required set of Service Definitions in the Root Device, and the hierarchy of required Devices and Service Definitions.
Service Definition. The formal definition of a Service Type. A Service Definition includes a Service Type Identifier, definition of the Service State Table (SST), definition of the Service Command Set, the Service Control Protocol (SCP) and Service Control Protocol Declaration (SCPD).
Device. In the context of the Device Model, a container for Services. A Device generally models a physical entity such as a VCR, but can also represent a logical entity. A PC emulating the traditional functions of a VCR would be an example of a logical device. Devices can contain other Devices. An example would be a TV/VCR packaged into a single physical unit. UPnP enables the association of user interface (display icon and root web page) with every Device, including Root Device.
Root Device. The topmost Device in a hierarchy of nested Devices. A Device with no nested Devices is always a Root Device.
Device Type. A relatively high level classification of Devices with common functionality. Device Type is intended to enable Devices to be simply and automatically grouped for presentation. An example of a Device Type is “VCR”. Device Types are formally defined in terms of a required set of Service Definitions of minimum version that a compliant Device must support. UPnP supports searches for all Devices of a specified Device Type.
Device Type Identifier. A unique identifier that identifies a Device Definition. This identifier adheres to the format of a Uniform Resource Identifier (URI). See, T. Berners-Lee, R. Fielding, L. Masinter, “Uniform Resource Identifiers (URI): Generic Syntax”, which can be found at http://www.ietf.org/rfc/rfc2396.txt.
Device Friendly Name. A human readable string that is initialized by vendors at the time of manufacturer of a Device. Every Device, including Root Devices, has a Device Friendly Name. A typical Device Friendly Name will contain manufacturer and model information, and is used to enable a more precise identification of a UPnP Device from the set of discovered Devices. Once identified, the Unique Device Name (UDN) can be used to unambiguously identify the same Device in the future. UPnP enables Device Friendly Names to be changed by User Control Points. The Device Friendly Name should not be used as device identifier.
Unique Device Name (UDN). The fundamental identifier of a Device. Every Device, including Root Devices, has exactly one UDN. The UDN is globally unique and permanent, even across power cycles and physical location changes. The UDN is the only UPnP device identifier guaranteed never to change. UPnP enables searches for devices by UDN.
Description Document. A structured unit of data that is used by a User Control Point or UPnP Bridge to learn the capabilities of a Controlled Device. Description Documents are retrieved from the Description Server on a UPnP Controlled Device. There is one Description Document for every Root Device that describes the Root Device and all non-Root Devices. Description Documents adhere to XML grammar. To support localization, multiple Description Documents can exist. A User Control Point requests the preferred localized Description Document by using the standard HTTP “accept-language” header.
Service. The fundamental UPnP controllable entity (but not the finest level of control). An example of a Service is “Clock”. Services are defined with a mandatory common base set of functionality. Vendors can extend the base set with proprietary extensions provided the base functionality is implemented. Service Definitions are versioned and later versions are constrained to be supersets of previous versions. UPnP enables searches for all Devices that contain a specified Service of a minimum version. This search would find all clocks, regardless of their packaging. A search for Device Type “Clock” would be used to find only stand-alone clocks.
Service Type. A classification of Services by their function.
Service Type Identifier. A unique identifier that identifies a Service Definition. This identifier adheres to the format of a Uniform Resource Identifier (URI). See, T. Berners-Lee, R. Fielding, L. Masinter, “Uniform Resource Identifiers (URI): Generic Syntax”, which can be found at http://www.ietf.org/rfc/rfc2396.txt.
Service State Table (SST). A logical table consisting of rows of [Variable, Type, Legal Values, Default Value, Current Value] that represents the current electrical, mechanical and/or logical state of a Service. SST instances are stored on the Controlled Device itself and are the ultimate authority of the state of the Service. All local user interface, such as front panels or wireless remotes are required to update the SST on UPnP compliant devices.
Service Command Set. A set of Commands that can be invoked on a Service. Commands generally result in changes in the Current Value field of one or more rows of a SST. Commands are logically represented in the format Command (Variable=New Value, Variable=New Value, . . . ). Services must accept or reject the complete set of changes to a SST. There is a mandatory standard Query Command that is used to retrieve the Current Value of any row of a SST.
Service Command Set Definition:
Service Control Protocol (SCP). The protocol used to invoke Commands against a Service and to return results. There is exactly one SCP per Service Definition. SCPs adhere to the grammar of SCP XML schema. SCPs can be generated by an automated tool that accepts a SST Definition and a Command Set Definition as input.
Service Control Protocol Declaration (SCPD). A formal representation of the schema of a Service. The SCPD declares the rows of a Service's SST and the associated Command Set. SCPDs are uploaded from Controlling Devices in their Description Documents and enable User Control Points or Bridges to invoke Commands on the Service without any prior or persistent knowledge of the capabilities (or schema) of the Service. There is exactly one SCPD per Service Definition. SCPDs adhere to XML grammar. SCPDs can be generated by an automated tool that accepts a SST Definition and a Command Set Definition as input.
Event. An unsolicited message generated by a Controlled Device and delivered to one or more User Control Points. Events are used to maintain a consistent view of the state of Service across all interested User Control Points. UPnP leverages the GENA event architecture (see “Generic Event Notification”) to transport event messages. All events are delivered using TCP/IP for reliability.
Generic Event Notification (GENA). An event transport protocol. GENA leverages TCP/HTTP as a transport. GENA has been submitted as an Internet Draft to the IETF. See, J. Cohen, S. Aggarwal, Y. Goland, “General Event Notification Architecture Base: Client to Arbiter”, which can be found at http://www.ietf.org/internet-drafts/draft-cohen-gena-client-00.txt.
Simple Service Discovery Protocol (SSDP). A simple network device discovery protocol. UPnP uses SSDP to allow User Control Points to find Controlled Devices and Services. SSDP operates in a default, completely automatic multicast UDP/IP based mode in addition to a server-based mode that uses TCP/IP for registrations and query. Transitions between the default dynamic mode and server-based mode are automatic and transparent. SSDP enables every Controlled Device to control the lifetime that its Description URL is cached in all User Control Points. This enables a Controlled Device to remain visible to User Control Points for a relatively long time (through power cycles), in addition to enabling a Controlled Device to appear and disappear very quickly, all under the control of the Controlled Device. SSDP and related Multicast and Unicast UDP HTTP Messages specifications have been submitted as Internet Drafts to the IETF. See, Y. Goland, “Multicast and Unicast UDP HTTP Messages”, which can be found at http://www.ietf.org/internet-drafts/draft-goland-http-udp-00.txt; and Y. Goland, T. Cai, P. Leach., Y. Gu, S. Albright, “Simple Service Discovery Protocol/1.0”, which can be found at http://www.ietf.org/internet-drafts/draft-cai-ssdp-v1-02.txt.
Client. In the context of UPnP, Client refers to a module that initiates a TCP/HTTP connection to a peer HTTP server.
Server. In the context of UPnP, Server refers to an HTTP server. This is a module that accepts incoming TCP/HTTP connections and either returns a web page or forwards the payload data to another module. Client and Server describe only the direction of initiation of TCP/HTTP connections. There is no relationship between the low level concepts of Client and Server and the high level concepts of User Control Point and Controlled Devices. Logically, User Control Points always discover and initiate communication with Controlled Devices, but this communication requires Client and Server functionality on both sides.
Hostname. A Hostname is the Domain Name System (DNS) or NetBIOS Name Service (NBNS) that, when resolved to an IP address, represents a network interface that can be used to establish TCP/IP level connectivity to User Control Points, Controlled Devices or Bridges. Hostnames can be used to provide persistent network level addressing on a network where IP addresses are dynamically assigned and of unknown lifespan or to integrate with an existing managed network. UPnP provides an algorithm for seeding a device's hostname from its UDN at manufacturing time.
Uniform Resource Locator (URL). A format for expressing web addresses. URLs minimally contain an identification of the protocol family that the URL is valid for, a Hostname, and a path. UPnP uses URLs as addresses whenever the module accepting the incoming connection is an HTTP server.
Description URL. The URL returned from a Controlled Device or Bridge in response to any UPnP SSDP query. This URL always points to a Description Server on the Controlled Device. An HTTP GET can be issued on this URL to retrieve the Description Document. This URL is valid as an address for the lifetime of the Hostname embedded in the URL.
Discovery Server. The module that runs in a Controlled Device or Bridge that responds to SSDP queries. This Server is unique in that it must support UDP/HTTP rather than just TCP/HTTP.
Discovery Client. The module that runs in a User Control Point that initiates SSDP queries.
Description Server. The module that runs in a Controlled Device or Bridge that responds to HTTP GETs and returns Description Documents. This service consists of a TCP/HTTP server than can retrieve and return a Description Document from persistent storage (like a filesystem).
Visual Navigation. User Control Point functionality that displays the icons of discovered Devices and enables the transfer of control to a browser or application to interact with the Controlled Device. In Windows, Visual Navigation could be implemented as a folder of icons.
Presentation URL. A URL that can be used by a User Control Point to navigate to the Presentation Server of a Controlled Device. This URL is returned in the Description Document and is valid as an address for the lifetime of the Hostname embedded in the URL. All Devices, including non-Root Devices, can have an associated Presentation URL.
Browser. The Presentation Client. A web browser extended with a Rehydrator.
Control URL. A URL that can be used by a User Control Point to navigate to the Control Server of a Controlled Device or Bridge. This URL is returned in the Description Document and is valid as an address for the lifetime of the Hostname embedded in the URL. All Services have an associated Control URL.
Control Server. The module that runs in a Controlled Device or Bridge that responds to Commands invoked on a Service by a User Control Point. Commands are encoded using the SCP specified in the Service Definition. This service consists of a TCP/HTTP server than passes control to the native control logic of a Service, updates the SST and generates an event if the SST changes.
Rehydrator. In UPnP, the Control Client. A User Control Point module that translates between native operating system APIs and SCPs and events. The Rehydrator uploads SCPDs from Controlled Devices and Bridges and generates appropriate SCPs in response to application API requests to invoke Commands.
Event Subscription URL. A URL that can be used by a User Control Point to navigate to the Event Subscription Server of a Controlled Device or Bridge. This URL is returned in the Description Document and is valid as an address for the lifetime of the Hostname embedded in the URL. All Services have an associated Event Subscription URL.
Event Subscription Server. The module that runs in a Controlled Device or Bridge that responds to GENA SUBSCRIBE requests from User Control Points. A SUBSCRIBE informs the Controlled Device or Bridge of the User Control Point's desire to receive future events. This service consists of a TCP/HTTP server that adds the User Control Point's Event Sink URL to the list of destinations to be NOTIFY'd whenever the SST associated with the Service changes.
Event Subscription Client. The module that runs in a User Control Point that sends GENA SUBSCIBE messages to the Event Subscription Server.
Event Sink URL. A URL, supplied by a User Control Point, that is used as an address to send event NOTIFYs to. This URL is valid as an address for the lifetime of the Hostname embedded in the URL. There is no explicit relationship between Event Sink URLs and Subscription Identifiers.
Subscription Identifier (SID). A header in the GENA NOTIFY message that identifies the source of an event. In UPnP, the SID can be considered as an alias for the Event Source instance.
Event Sink. The module that runs in a User Control Point that accepts incoming GENA event NOTIFYs. This service consists of a TCP/HTTP server that passes the event information to interested applications running on the User Control Point.
Event Source. The module that runs in a Controlled Device or Bridge that sends GENA NOTIFYs to the Event Sink Servers of SUBSCRIBES User Control Points.
Domain Name System (DNS). A distributed system of servers that locates the IP addresses of other computers on a network based on their hierarchical names.
NetBIOS Name Server (NBNS). A server that locates the IP addresses of other computers on a network based on their flat NetBIOS computer names.
Multicast DNS (MDNS). A peer-to-peer translation scheme that does not require involvement of DNS servers.
UPnP Technologies Overview
An overview of technologies utilized in UPnP follows.
Device Discovery: Simple Service Discovery Protocol (SSDP)
TCP/IP provides the ability to initiate a connection with a specified application running on a specific device, provided both the network address of the device (IP address) and the application address (port) are known. Generally, application addresses (ports) are standardized and widely known, but the problem of learning the IP address of a device remains.
Simple Service Discovery Protocol (SSDP) is a protocol that enables devices to learn of the existence of potential peer devices and the required information (an IP address) needed to establish TCP/IP connections to them. The successful result of an SSDP search is a Uniform Resource Locator (URL). The Hostname embedded in the URL can be resolved to an IP address that can be used to make a connection to the discovered device. The name to address resolution is outside of the functionality of SSDP.
SSDP specifies a default, completely automatic, best-effort multicast UDP-based operating mode, in addition to a server mode that uses TCP for registration and query. Fall-forward to server mode and fallback to the default dynamic mode can occur automatically and transparently as a server is added or removed from a network. Server mode can be used to reduce network traffic, to implement searches based on location or policy and to integrate with a directory system.
SSDP requires that all devices specify a maximum lifetime that SSDP level knowledge of the device will remain cached in other network devices. If a device does not refresh the cache of other network devices before this interval expires, the device will disappear from the network. This interval can be chosen to be larger than a typical power down cycle to enable device visibility to persist for a relatively long time, or a smaller interval can be chosen to enable more dynamic visibility control. In all cases, devices that are abruptly removed from the network will eventually disappear from all networked devices.
In response to an SSDP search, UPnP devices return a Description URL in the SSDP Location and optionally the Alternate Location (AL) SSDP headers. An example location header is a follows:
In this example, the device.local is the Hostname of the Controlled Device, and the “description/path/description.xml” element of the URL is the path and name of the Description Document on the device.
Eventing: Generic Eventing Notification (GENA)
Eventing, in the context of UPnP, is the ability for a device to initiate a connection at any time to one or more devices that have expressed a desire to receive events from the source device. Events are used to enable synchronization among multiple devices organized into a many to one relationship. UPnP events are mainly used for asynchronous notifications of state changes.
TCP/IP provides the fundamental support for the connections that carry event information. Generic Event Notification (GENA) adds conventions for establishing relationships between interested devices and an addressing scheme to enable the unambiguous delivery of events. GENA leverages HTTP addressing and encapsulation.
User Control Points, Controlled Devices and Bridges
With reference now to
The primary distinction between a User Control Point 104-105 and a Controlled Device 106-107 is that the User Control Point is always the communication initiator. After the initial communication, User Control Points can receive events from Controlled Devices.
Controlled Devices 106-107 are responsible for storing the state of Services. User Control Points are required to synchronize to the state on Controlled Devices and to share state directly among themselves.
User Control Points typically have user interface that is used to access one or more Controlled Devices on the network. Controlled Devices only have local user interfaces.
Bridges 120 (
The following table lists the modules in the User Control Points 104-105 and Controlled Devices 106-107, along with their functions.
The UPnP Device Model 200 shown in
The fundamental controllable entity in UPnP is a Service 210-217. Every running instance of a Service includes:
The SST 230 can be used to represent the operational mode of device or to act as an information source or sink for structured data or simple files. The SST of a VCR 254 (
The logical structure of a SST published in the Service Definition, but the actual storage format of an instance of a SST is entirely up the device. The only interaction with a SST is through a formal application level network protocol.
The formal definition of a Service (Service Definition) includes:
According to the device model 200 shown in
The UPnP Device Model was designed to be general and flexible. It should be possible to model an entire Nuclear Power Plant as a single Service or as a deeply nested hierarchy of Devices and Services. In general, a Service 210-217 is cohesive set of functions that enables flexible packaging into a variety of Devices. Services can be versioned independently of Devices.
All Devices, including Root Devices belong to one or more Device Types. Device Types are intended to enable instances of Devices to be simply and automatically grouped for presentation. An example of a Device Type is “VCR” 254 (
UPnP enables SSDP level searches for a unique instance of a Device (by UDN), all Devices of type Device Type and all Devices that contain at least one Service Type of minimum version. The result of an SSDP search is always a URL that points to the Description Document contained in the Root Device. In the event that matching Device is not the Root Device, the Description Document has a tree of nested Devices that can be traversed to find the matching Device.
Every Device includes:
Every Root Device 202 also includes the Description Document 226 and Description Server 228 for all Devices under and including itself.
The formal definition of a Device (Device Definition 226) includes:
There can be many Device Definitions that belong to a single Device Type.
The formal definition of a Device Type includes:
Service State Table
A Service State Table (SST) logically consists of rows of:
Variable, Type, Legal Values, Default Value, Current Value Although entries of the Service State Table in UPnP consist of these five items, the state table alternatively can contain fewer or additional items. Generally, each entry will minimally consist of a Variable name or identifier, and its current value.
The following table lists various Types available in UPnP.
The ByteBlock is essentially a data buffer. In one use, a variable of this type can be used to effect transfer of a file from the Controlled Device to the User Control Point. The file to be transferred is kept in the Service State Table as the current value of this variable. On a change in the file, the file is transferred to any subscribing User Control Point in an event notification.
The reason for representing Services this way is to ensure that the state of a Service is easily available in a common way to multiple User Control Points.
An SST can be used to represent to current operational mode of device, act as an information source or sink and/or simply be a repository for commands. The SST of a VCR Service could represent the current transport mode, tuner channel selection, input and output switch selections, audio and video decoding format and current timer program. Alternatively, the VCR 254 could be represented as a Transport Service 260, Tuner Service, I/O Switch Service, A/V Decoding Configuration Service and Programmable Timer Service 261.
The SST of a clock 251 would likely represent the current time. Additionally an alarm clock could include Service Variables to configure the clock.
The SST of an image rendering device could implement a video frame-buffer that can accept raw pixel information or formatted JPG files. The SST of an audio or video playback device could implement a transfer buffer or queue of material to be played. The SST of PDA could implement a collection of formatted data that has changed and needed to be synchronized with another device, in addition to a transfer buffer for accepting incoming formatted data.
User Control Point Synchronization
In accordance with an device state and eventing model illustrated in
With reference now to
While UPnP modules are described as HTTP servers, there is no requirement that implementations be based on actual web servers. In most cases, the job of the HTTP server is simply to accept the incoming connection, look at the local destination part of the address (the path) and forward the payload to another module. UPnP enables, but does not require, that all HTTP Servers be based on a common software implementation or runtime instance. Controlled Devices and Bridges can include a TCP port specification as part of a URL to override the default value of 80.
The successful result of a UPnP SSDP level search is always one or more Description URLs. These URLs can be used to navigate to the Description Document of a Controlled Device or Bridge. A User Control Point uploads the Description Document and extracts the URLs of the Servers running on the Controlled Device or Bridge.
All URLs returned in the Description Document have a lifetime equal to the lifetime of the Hostname embedded in them. User Control Points can store these URLs as addresses without going through a search sequence first. Once they have been advertised in a Description Document, Controlled Device and Bridges cannot arbitrarily change Server URLs.
Whenever a Hostname changes, all URLs associated with all Devices addressed by that Hostname are invalidated. The UDN is the only UPnP identifier guaranteed never to change. Any persistent associations maintained by applications should at least store the UDN to able to unambiguously identify the target Device.
The lifetime of a Description URL is determined by Controlled Device or Bridge that advertises it. If a Controlled Device or Bridge allows an SSDP advertisement of a Description URL to expire, the URL is invalidated.
User Control Points use the Event Subscription URL returned by the Controlled Device or Bridge to connect to the Event Subscription Server. This server does the housekeeping of remembering all User Control Points that are interested in receiving Events on a Service. The Event Subscription Server needs an address to send the events back to. This address is called the Event Sink URL, and is supplied to the Controlled Device or Bridge in the GENA SUBSCRIBE message. The lifetime of an event subscription, and the Event Sink URL, is determined by the timeout on the SUBSCRIBE message.
Further details of UPnP addressing are listed in the following table.
Device Discovery and Identification
UPnP enables SSDP searches for a unique Root or non-Root Device by UDN, devices of a specified Device Type and devices containing a Service of a specified Service Type.
SSDP specifies Service Type (ST), Notification type (NT), and Unique Service Name (USN) header fields for queries and for announcements. UPnP uses the ST or NT header to carry one of the UPnP defined identifiers. A unique USN is required for each unique SSDP announcement.
Multiple instances of the same Service Type within a Controlled Device 106-107 or Bridge 120 are not independently announced.
UPnP search identifiers are used during the discovery process. The result of a successful discovery is one or more Description URLs. The format for search identifiers is:
SSDP specifies that SSDP announcements must be made for all SSDP searchable values. The SSDP announcements with “all” as the notification header value must carry the Root Device UDN as the USN header value. SSDP announcements for Device Types must carry the UDN of the Root Device concatenated with the Device Type URI as the USN header value. SSDP announcements for a Service Type will carry the UDN of the Root Device concatenated with the Service Type URI value as the USN header value. SSDP announcements of UDNs will repeat the UDN value as the USN header.
UPnP Bridges 120 (
The UPnP Description Document 226 (
The Description Document is an XML document. UPnP defines the use of HTTP and XML for the Description Document and wire protocols. UPnP adheres to the schema declaration rules of XML-Data and Y. Goland, “Flexible XML Processing Profile.”
The top level XML elements are separated into three categories: per Device, per Service and shared.
With reference now to
User Control Points 104 are not required to have any prior knowledge of the SCPs 402 required to control the Services on the various devices. Therefore, a Controlled Device or Bridge must be able to describe to a User Control Point the protocols required to control its Services, such that the User Control Point will be able to implement these protocols dynamically. This requires a standard way of declaring Service Control Protocols in a concise and unambiguous fashion. UPnP introduces a technique for declaring Service Control Protocols using a series of XML documents.
A Rehydrator 410 is a module that exposes a suitable API to applications and either invokes Commands on a Service or queries the state of that Service, or receives and responds to events. The primary job of the Rehydrator is to map between API calls and the Service Control Protocol sequence that invokes the Command.
As part of the Service Definition 406, a Service State Table 230 and Command Set 408 are defined. These things can be combined in a deterministic way defined by UPnP to produce a Service Control Protocol Definition (SCPD) 406, which includes a Service Control Declaration 404 and a Service Control Protocol 402. The SCPD 406 is a representation of the schema of a Service. It is possible to reconstruct the SST, Command Set and SCP from the SCPD.
The SCPD is directly embedded into the Description Document 226 of a Controlled Device. When the Description Document is uploaded into the User Control Point 104, the Rehydrator 410 can extract the SCPD from it. At this point, the Rehydrator has enough information to issue Service specific SCPs 402.
General Operation of the Rehydrator
More generally with reference to
The Description Document 226 also includes a declaration of the methods, properties and events for the Service. Based on this declaration, the Rehydrator produces a corresponding programmatic interface for use by applications at the User Control Point. The programmatic interface is an application programming interface that can be in the form of an object integration interface of an object-oriented programming model, such as Microsoft COM, CORBA, Java classes, and scripting engine name extensions. In the example illustrated in
Accordingly, the Rehydrator operates as a universal proxy object with data-driven conversion of programmatic interfaces to network data messages. Further, the Rehydrator produces the programmatic interface at the User Control Point based solely on an XML data description. This operation allows the Rehydrator to produce just-in-time transient interfaces to remote device Services without the complexity of code downloads and installation or configuration. Upon a later release of the interface by the application, the Rehydrator destroys the interface without need to de-install or clean up persistent configuration data in a registry or configuration file of the operating system or object execution run-time.
Summary. With reference to
Basic Functionality. The preferred implementation of the Rehydrator is able to translate a service control call to the UPnP API into the appropriate network messages defined by the Service Control Protocol.
Asynchronous Event Notification. The preferred implementation of the Rehydrator is able to notify UPnP API clients of any asynchronous events generated by the devices they are controlling. Event notification is done by means of the event interfaces defined below.
Error Reporting. For a variety of reasons, state variable queries and action invocations may fail. The preferred implementation of the Rehydrator is able to provide a way to communicate the success or failure status of such operations to the parties initiating them.
Rehydrator Implementation Design. As illustrated in
Creating Service Objects. When the Device Finder 450 creates a Device object, it invokes the Rehydrator 410 to create Service objects 460 for each of the service instances on that device. Each service instance supports a particular Service Control Protocol and the Rehydrator needs a description of this protocol in order to create a properly hydrated Service object.
The Service Control Protocol is declared in two separate XML documents: the DCPD and the Contract. The Rehydrator needs the information in both documents. These two documents are passed to the Rehydrator as IXMLDOMDocument interface pointers in the RehydratorCreateServiceObject( ) API call.
The Rehydrator stores the information that is specific to a particular service instance as private data within the Service object itself. This includes the control URL and information about the control server 232 (such as the HTTP verbs it supports). The Service Type Identifier is the link between the Service object that represents one instance of a service type and the Rehydrator internal data structures that contain information common to all instances of that service type. The Service Type Identifier is stored as a private data member in the Service object.
Querying Service Properties. Applications can query the values of service properties by invoking the IUPnPService::GetProperty( ) method on a Service object. Internally, this method makes a call to the RehydratorQueryState Variable( ) function.
The first two in parameters to this function supply the service instance specific information: the HTTP verb to use and the control URL to which the network messages will be targeted. The third parameter is the Service Type Identifier that will be used to locate the Service Control Protocol information in the Rehydrator's internal data structures. The fourth parameter is the name of the variable that is being queried (the Rehydrator will validate this against is internal list of state variables exported by the service) and the final parameter is the address of a VARIANT structure in which the Rehydrator will place the variable's value.
This function will generate an HTTP request to the control server on the device. The body of this request will be an XML fragment containing a XOAP-encoded request for the variable's value. The following is an example of such a request (the exact header and payload format of this message is defined in the service contract):
The control server will respond to this message with another XML fragment: the XOAP-encoded method response. The following is an example of such a response:
The rehydrator will extract the return value from this XML fragment, place it in the VARIANT structure whose address was passed as the last parameter to RehydratorGetServiceProperty( ) and then return.
Invoking Service Actions. The process of invoking a service action is very similar to querying a state variable. An application calls IUPnPService::InvokeAction( ) on a Service object, passing it the name of an action to invoke, and an array of arguments to the action. Internally, IUPnPService::InvokeAction( ) calls RehydratorlnvokeServiceAction( ), declared as shown below.
As was the case for querying state variables, the service instance specific information is passed in the first two parameters, followed by the Service Type Identifier in the third. The action name and an array of arguments are passed as the next two parameters, and the final parameter is the address of a variable in which to store the status of the operation.
RehydratorInvokeServiceAction( ) will send an HTTP request to the control server identified by the second parameter. As before, the body of this message will be an XML fragment containing a XOAP-encoded method call. An example HTTP request to invoke an action is shown below.
The encoding of the body of this message is again specified in the service contract. The Rehydrator will wait for the HTTP response to this request, which would look something like the example below.
After receiving a response such as this, the Rehydrator will extract the return value, place it in the out parameter it was passed, and then return.
With reference to
The HTTP GET must include an accept-language header that is used to request the preferred language of the response. If the requested language is not supported, a Description Document in the default language supported by the Controlled Device or Bridge may be returned.
An HTTP GET is used to retrieve sub elements of a Description Document that are expressed as URLs.
URLs embedded in Description Documents 226 take one of 3 forms: a fully qualified URL or a relative URL.
Fully qualified URLs take the form:
The devicename part of the URL is a Hostname or IP address and the pathname is a filesystem path or equivalent. A fully qualified URL is used “as is” to establish an HTTP connection to a device.
A relative URL does not contain the “:” character and is of the form:
Relative URLS are a compact representation of the location of a resource relative to an absolute base URL. All relative URLs in a Description Document are appended to the value of the Description Document element <URLbase> to form fully qualified URLs.
Some elements of a Description Document are binary. XML does not directly support the embedding of binary data. In order to include binary data directly in a Description Document, one must convert the data to text using the Base 64 encoding scheme. This tends to increase the size of the data by 25% on the average. Much of this overhead can be eliminated if the binary data is passed by reference instead of by value. To reference binary data, a URL to the data is provided in a Description Document. The binary data can be retrieved by doing a HTTP GET with that URL.
As an example, consider the <image> element in the following Description Document:
The icon would be retrieved with an HTTP GET of the following format:
The HTTP response would look like:
Description Document Layout
The basic layout of the Description Document 226 is shown in
The following table lists Description Document elements that are sub-elements to the root element.
The Description Document elements listed in the following table are associated with devices.
The following elements of the Description Document are associated with Services.
Service Control Protocol and SCP Declaration
As part of the Service Definition 406 shown in
The declaration of such a protocol must specify the list of Variables that can be queried, the set of Commands that can be invoked, as well as the wire protocol (the content and sequence of network messages) required to carry out these operations. SCPD is specified in two XML documents. The first or Service Control Definition document 404, written in a language called Service Control Protocol Declaration Language (SCPDL), declares the list of state Variables and Commands associated with the Service Type to be controlled by the protocol. The second or Service Control Protocol document 402 is written in Contract Definition Language (CDL) and declares the wire protocol that will be used to query the values of the state variables and invoke the actions associated with the service.
Declaring the Service State Table and Command Set
A SCPDL document 404 is used to specify the list of state Variables that a SCP can query and the set of Commands that it can invoke. SCPDL is an XML schema, a set of rules for writing XML documents (Service Control Protocol Declarations).
The <actionList> element contains an <action> element for every action associated with the Service. The elements within an <action> element specify the name of the action and any arguments the action may take. In this case, the service supports two actions that do not take arguments, ChannelUp and ChannelDown, and another, SetChannel, that takes a new channel number as an argument. The <argument> element and the elements nested within it define the argument. The <relatedStateVariable> element within <argument> specifies the name of one of the state variables to which the argument is related. In the UPnP Device Model, all arguments to actions must correspond directly to some state variable.
Declaring the Contract
The Contract is a specification of the wire protocol that will be used to query state Variables, invoke Commands and carry notifications or events. This contract specifies the type of protocol used, the network endpoint to which messages are sent, the contents of those messages, the contents of the expected responses and the contents of events. Contracts are written in Contract Definition Language (CDL).
All UPnP SCPs will use essentially the same contract. A specific contract applies to a single Service instance (since it specifies the network endpoint to which messages are sent and network endpoints are specific to service instances). However, other than the network endpoint definition, all contracts for all Service instances should be the same.
Basic UPnP Eventing Architecture
With reference to
What is an Event?
Property change events are defined as any change in the value of a row of the Device State Table (DST) 230 (
What is a Notification?
A UPnP event notification is an XML message sent over HTTP/TCP to each and every subscriber to a particular UPnP service. The content of the XML is defined below. The important contents of this message are the unique identifier for the subscription, the property name, new value, and property type.
In UPnP, the listener to Notifications is the SSDP service itself. SSDP already listens on another multicast address for “alive” and “byebye” messages sent by UPnP devices. The same listener will listen on a TCP port for notifications sent. All subscriptions sent from that UCP contain the same callback URL and so all notifications will be directed to that URL. When a notification arrives the SSDP service will examine the NT header of the message and determine if it is an event notification. If so, the message is parsed further to determine if it should be forwarded on to subscribers (which must exist). GENA defines the format of the HTTP message, what headers can be used, and what they can be used for.
GENA is the protocol of communication that, in a preferred embodiment, UPnP devices use to send event notifications. Therefore, UPnP devices that wish to notify UCPs of state changes are recommended to use GENA. Notification subscribers will never be required to interact with a UPnP device directly and so they are not required to use GENA. The eventing API will encapsulate this complexity. Other appropriate event transport protocols may be used, such as publish/subscribe systems.
Applications written in C (C Application 604) will be able to utilize the SSDP C API 610 to receive callbacks when notifications are processed by the SSDP service. This is analogous to SSDP clients registering for notifications that services have become available. When a UCP registers for a notification, it passes as a parameter the URL of the service for which it is interested in receiving notifications. This URL is obtained from the description document for that service. (When a service is registered on a UPnP device, it uses this same URL to listen for subscription requests).
When a notification message is received by the SSDP service listener, the SID header is checked against the list of subscribers it maintains. If a subscriber is found, the callback function for that subscriber is invoked, with one of the parameters being the contents of the notification message. The notification client that implements the callback function can process this message in any appropriate way.
Notifications in the UPnP API
The UPnP API 410 is a consumer of the basic C interface provided by the SSDP C API 610 component. In order to integrate seamlessly, the registration of notifications is handled by the Service Object 612 inside the UPnP Object Model. Service objects will register for notifications when they are created. This ensures that the DST is maintained by the UPnP API and is kept up to date. They will implement the callback function required by the registration function. If this callback function is invoked, it will pass on that notification to UCPs. The UCPs can be written in C, C++, VB, or script code, so the mechanism for passing on notifications can be different.
Eventing Subsystem Terminology
UCP—User control point. Any piece of software that searches for devices and controls them.
CD—controlled device. A hardware or software device that announces its availability thru SSDP and allows control by UCPs.
Subscriber—A UCP who wishes to be notified of event changes.
Notifying Resource (or simply “Resource”)—For the purposes of this document, this will always be a service contained within a UPnP CD 106.
Event Source—a service that provides events. UPnP services are event sources. All notifying resources are event sources and vice versa.
Event—message generated when a change in a resource's state occurs.
Property—a single entry in the service's state table whose DefaultValue can change. Properties and events always have a one to one correspondence.
Subscribing to Resources
Integrating with the UPnP API
The UPnP API 410 exposes several interfaces with which a consumer can find and enumerate devices, control services, and get properties on devices and services. To allow the integration of events into this model, we add a new property to the IUPnPService interface called EventHandler. When this property is set, it tells the Service object 612 that its client is interested in receiving notifications for that service. The SSDP API RegisterNotification( ) API is called when the Service object is created so that it can maintain a local copy of the DST for that service. The Service object knows the URL of the service and therefore it can provide this as a parameter to RegisterNotification( ). RegisterNotification( ) is also provided a callback function which is a static member of the Service object class. This function will be invoked for each and every notification sent by that particular UPnP service.
The Notification Callback
The Service object 612 includes a static member function called EventNotifyCallback( ) which is invoked for each notification sent by the UPnP service. The callback is passed the entire HTTP message contents in a structure which is a parameter to the function. The prototype looks like this:
The ssdpType parameter should always be SSDP_PROPCHANGE. The pssdpMsg parameter contains the relevant information about the event. The key piece of information is the body of the XML message. The body contains information about what property changed, what its new value is and what type it is, among other information. The pContext parameter will always be the this pointer of the Service object. This allows the code to call a method to fire the event to the UCP. The callback will parse the XML body using the XML DOM services. Property changes are iterated and the local DST is updated to reflect these changes. After this processing is done, an event notification may be fired for each property that was changed to the owner of the subscription if one exists. Depending on what environment the owner is written in (C++ or script, etc . . . ), a different mechanism for firing the event may be employed.
A special case for this process is the very first notification received after a subscription is established. This notification contains the entire set of properties and their values and is used to locally sync up the DST. Events will not be fired to clients of the UPnP API in this case.
When the EventNotifyCallback( ) function is called, the local copy of the DST for the service is updated. After this, an event needs to be fired if a subscriber exists. A subscriber exists if the put_EventHandler( ) method was called, either from VBScript, C++ code, or another source. To abstract away this complexity, a new interface called IUPnPEvents is needed.
This interface currently has one method called NotifyEvent( ) which takes several parameters (TBS). When put_EventHandler( ) function is called, its argument is an IUnknown. This pointer is Queryinterface'd( ) for IDispatch first, and if it succeeds, then IDispatch::Invoke( ) is called with DISPID 0 to invoke the default method. This allows VBScript 602 to be called. If that fails, however, it is Queried for IUPnPEvents, and if that succeeds, the NotifyEvent( ) method is called with the same parameters as for Invoke( ). The handles C++ UCPs effectively.
Subscribing with C++
To subscribe to a UPnP service from C++, a UCP instantiates a UPnP service object, issues Queryinterface( ) to it for IUPnPEvents, and calls the IUPnPEvents::SetEventCallback( ) function. This function takes 2 parameters, a callback function pointer and a context pointer.
Subscribing with VBScript
To subscribe to a UPnP service's events, all that needs to be done by a script 602 is to create a function or subroutine as a handler function and set the pointer of that function to the EventHandler property of the Service object. Now, anytime an event is fired, this VBScript function or subroutine will be called. In VBScript, this is written as the following:
In this example, the script enumerates all devices, looking for any device that supports the “Clock” interface. When it finds a device that supports that interface, it enumerates that device's services looking for the one that has the “clock.v1” interface. Once it finds that service, it sets that service's EventHandler property to the VBScript subroutine called “clock_PropertyChanged”. This name is arbitrary.
Sending and Receiving Notifications
GENA Client API
GENA clients are actually UPnP services. A GENA client creates a new event source when it is initialized. The GENA client API 620 facilitates this. It also provides a way for GENA clients to send their notification messages. It is also important to note that the HTTP server that lives on the UPnP device is also a client of this API. The GENA client API consists of the following functions:
The RegisterUpnpEventSource( ) API gives a GENA client the ability to register itself as an event source. The prototype is as follows:
Parameters: szRequestUri [in] an arbitrary Request-Uri that SUBSCRIBE requests will be sent to. When a SUBSCRIBE request arrives at the given URI, it is acknowledged and the subscriber is added to the list of notification recipients. Note that this URI should match the URI provided in the description for this service. CProps [in] the number of properties that this event source provides. RgProps [in] Array of UPNP_PROPERTY structures which contain information about each property. The property information is derived from the DST for the event source.
Return Value: The function returns a TRUE if successful. If the given URL has already been registered as an event source, the return value is FALSE and GetLastError( ) returns ERROR_ALREADY_EXISTS.
Notes: The initial state of the event source needs to be given to the API so that it can effectively maintain the up-to-date state of the event source.
The DeRegisterUpnpEventSource( ) API gives a GENA client the ability to deregister itself as an event source. The prototype is as follows:
Parameters: szRequestUri [in] an arbitrary Request-Uri that SUBSCRIBE requests will be sent to. When a SUBSCRIBE request arrives at the given URI, it is acknowledged and the subscriber is added to the list of notification recipients. Note that this URI should match the URI provided in the description for this service.
Where szname is the name of the property, szValue is the current value of property, and szType is the type of property (string, integer, etc . . . ).
The SubmitUpnpPropertyEvent( ) API allows the GENA client to submit a UPnP property change event to be sent to subscribers as a notification. The prototype is as follows:
Parameters: “szRequestUri [in]” identifies the event source to which this event belongs. This is the same Request-Uri passed to RegisterUpnpEventSource( ). “DwFlags [in]” is unused. “CProps [in]” is the number of events that are being submitted. “RgProps [in]” is an array of UPNP_PROPERTY structures which contain information about each event.
Return Value: If the function fails, the return value is FALSE. The get extended error information, call the GetLastErroro function.
Notes: When a series of properties is submitted for event notification, the local version of the property state for the given event source is updated with the list of properties passed in. SubmitUpnpPropertyEvent( ) calls SubmitEvent( ) after it has generated an XML body.
The SubmitEventO API allows the GENA client to submit an unstructured event to be sent to subscribers as a notification. The prototype is as follows:
Parameters: SzRequestUri [in] identifies the event source to which this event belongs. This is the same Request-Uri passed to RegisterUpnpEventSource( ). DwFlags [in] Unused. SzHeaders [in] null-terminated text string containing the headers for the event, each separated by CRLF. SzEventBody [in] null-terminated text string containing the body of the event message
Return Value: If the function fails, the return value is FALSE. The get extended error information, call the GetLastError( ) function.
Notes: If no subscribers exist, the function does nothing. If one or more subscribers exist, a message is sent to each subscriber. SubmitEvent( ) will always send to all subscribers.
UPnP Controlled Device Architecture
In UPnP, every UPnP service 210-211 that supports property change event notifications is to be a GENA client. Therefore, when the service is initialized, it must register itself as a GENA event source. It will do this with the RegisterUpnpEventSource( ) API. This returns a handle which can be used in subsequent APIs.
RegisterUpnpEventSource( ) takes a URL and an array of properties as parameters. Inside the API, an entry in an array of structures is initialized and the index is returned as the handle. The structure contains the source URL as one of the members. A second member of the structure, an array of destination URLs, is left uninitialized. This is filled in each time as subscriber is added for that event source. Another member of the structure is the list of properties that this event source provides. This is effectively a cached copy of the DST for the event source. As events are submitted, the local properties are updated.
When SubmitUpnpPropertyEvent( ) is called, each property submitted replaces the corresponding property already maintained by the API. If no subscribers exist, the request to submit an event is ignored. If one or more subscribers exist, their callback URLs are looked up in the list of subscribers for the given event source and a NOTIFY message is constructed and sent to each URL, one at a time, in order of subscription.
If an event is submitted and no response is received (or a CD-side error occurs), the CD continues to attempt to send to the UCP. If the subscription timeout expires, then the subscription is removed. If the UCP becomes available again, it will re-subscribe because it will notice the sequence numbers are not contiguous.
When an HTTP server 626 receives a SUBSCRIBE message, it passes it along to a function which parses the message for the necessary information. The Request-URI identifies the service that is to be subscribed to. The callback URL is obtained from the “Callback” header. Since the Callback header can contain multiple URLs, it picks the first “http://” URL it finds. It then adds the subscriber to the list of subscribers for this event source. A unique subscription identifier is constructed which it will send back to the subscriber in the HTTP response to the SUBSCRIBE request.
If no event source matches the Request-URI from the subscription message, the HTTP server should return “404 Not Found”.
When a subscription is added, the local copy of the DST is sent as a NOTIFY message. This special NOTIFY message contains sequence number 0 which informs the UCP that this is an initial state population event and not a notification where every event has changed.
When a CD receives an UNSUBSCRIBE message, it checks the “SID” header to obtain the subscription identifier. It looks up the subscriber ID in the list of subscribers for that event source and removes the destination URL entry associated with it.
GENA Server API
GENA servers 630 are generally going to be UPnP UCPs. A GENA server is anything that receives and processes NOTIFY messages to handle notifications from resources and sends SUBSCRIBE and UNSUBSCRIBE messages to receive notifications from resources. These APIs leverage the already existing SSDP APIs. The following are the changes to the APIs:
The RegisterNotification( ) allows a UPnP UCP to request notification when an event occurs for a given UPnP service. The prototype is as follows:
Parameters: Nt [in] An enumeration that determines the type of notification requested. The values are: SSDP_ALIVE—a service has become available, and SSDP_PROPCHANGE—a property has changed on the service. SzResourceType [in] A null-terminated string specifying the resource type desired. For SSDP_ALIVE, this is the service type, for SSDP_PROPCHANGE this is unused. SzEventUrl [in] A null-terminated string specifying the URL that a subscription request should be sent to. FnCallback [in] A pointer to a function that will be called each time a notification is received. The function pointer is defined in the SSDP spec. PContext [in] This parameter is included as a parameter when invoking the client-supplied callback function.
Return Value: If the function succeeds, the return value is a handle used in a subsequent call to the DeregisterEventNotification( ) function. If the function fails, the return value is INVALID_HANDLE_VALUE error code. To get extended error information, call GetLastError.
UPnP UCP Architecture
When a UPnP UCP wishes to subscribe to notifications for a particular UPnP service, it calls the RegisterNotification( ) API. It passes to this API a notification type that identifies the type of notification being requested, a URL to which a subscription should be sent, and a callback function and context for use when the notification is received.
RegisterNotification( ) will compose a SUBSCRIBE message, using the data passed in, and send that to the URL specified by the caller. The Callback header of the SUBSCRIBE message will be composed on the fly, as an arbitrary URL for notifications to be sent to for this subscription. This callback URL will likely be a constant since the server API will always know how to handle requests sent to this URL. It will then send the SUBSCRIBE message and await a response.
RegisterNotification( ) in the SSDP API does not currently send HTTP requests, but it can be modified to do so. It also needs to await a response which it will also be modified to do so.
When the response is received, the Subscription-ID header contains a SID which is associated with the callback function specified by the caller.
Immediately after the response is received, the UCP should expect an initial NOTIFY message that contains the complete set of properties maintained by the CD. This becomes the local cached DST on the UCP side. From this point on, all modifications to the table are made via NOTIFY messages. This initial NOTIFY message will have sequence number 0 that indicates it is an initial property set and not an update. The UCP can use this information in any way it sees fit. This ensures the UCP's state table is always in sync with the one on the CD.
When a message is received by the HTTP server on the UPnP UCP, it is passed to a function which determines the method name and Request-URI. If this is a NOTIFY message, the headers are parsed and packaged up into a structure. The callback function that was specified to RegisterNotification( ) is called with that structure as one of the parameters. UCPs who implement the callback function can find the headers and body of the NOTIFY message and do additional processing based on the notification type.
This all requires that the SSDP HTTP server listen on a TCP socket in addition to the UDP multicast port it already listens to. However, once a NOTIFY message is received, it is processed in the same way regardless of from which connection it originated.
The following are subscription/notification failures that can occur and their solutions:
To protect against subscriptions that exist on the controlled device, but no longer on the UCP, we institute the timeout feature of GENA subscriptions. The scenario is this: A UCP subscribes to a CD, then the UCP reboots. Meanwhile, the CD is still trying to send notifications to that UCP. If the UCP never comes back, the subscription would be leaked because the UCP never told the CD that it was going away. So to correct this, each subscription request includes an arbitrary timeout value which indicates to the CD that the UCP will be re-subscribing every n seconds indicated in the timeout header of the subscription request. If the timeout expires on the CD, the subscription is removed. The UCP is required to re-subscribe before the timeout period has elapsed. If it fails to do so, the subscription will be terminated by the CD.
Some time before the timeout expires on the UCP, a re-subscribe message should be sent. The re-subscribe message is similar to the subscribe message, but it does not contain an NT or Callback header. If the UCP is unable to re-subscribe within the timeout period, the subscription will be terminated by the CD. If the UCP sends a re-subscribe after the CD has terminated the subscription, the CD will return “412 Precondition Failed”.
Reboot of a Controlled Device
If a controlled device reboots, information about all of its subscribers would be lost. To prevent this, the subscriber information will be persisted across reboots of the device. Because the subscription info contains a timeout member, the absolute expiration time will be used when the subscription information is persisted. That way, when the device comes back up, it can check the timeout for each subscriber and if that time has passed, the subscription will be removed.
Network Error Sending Event Notifications
If a controlled device receives an error sending an event notification to a subscriber, it will NOT cease to send notifications. It will continue to send notifications and receive errors until the subscription expires. The problem for the UCP is that it will have missed a number of event notifications and so its state table will be out of sync. To correct this, each event notification message will contain a 32-bit sequence number that starts at 0 and increments for each message sent to a subscriber. If a subscriber receives a notification with a sequence number that is not exactly one more than the previous notification, it will know that it has lost events and will ignore all future notifications until it receives one with sequence number 0 again. Events with sequence number 0 indicate that the event is an “initial state” event.
Once it realizes that is has lost one or more events, the UCP will send an UNSUBSCRIBE message, followed by a SUBSCRIBE message. This is not the same as a re-subscription because re-subscriptions do not cause the CD to start the sequence over at 0. In this case, the active unsubscribe/subscribe will cause the CD to restart the sequence at 0 and send the entire state table with the first notification message.
The SUBSCRIBE Message
When a UPnP UCP wishes to subscribe to event notifications for a UPnP service 210-211, it will form a SUBSCRIBE message of the following format:
The response is as follows::
This example of a GENA SUBSCRIBE request and response demonstrates a subscription to event notifications for “service1.” The host is “vcr.local.” All notifications for this service will be sent to the callback URL http://danielwe/upnp:923. In the response, the “Subscription-ID” header provides the subscriber with an identifier to use when it wants to unsubscribe to this resource. The “Timeout” header indicates that the subscriber will send a re-subscription request before 10 minutes have elapsed. If the device does not receive this request within that period of time, it will remove the subscription.
The Re-SUBSCRIBE Message
When a UPnP UCP wishes to re-subscribe to event notifications for a UPnP service, it will form a SUBSCRIBE message of the following format:
The response would be as follows::
Note that the NT and Callback headers are absent, but the SID header exists. This tells the CD 106 which subscription is being renewed and restarts the timeout. When the CD receives this message, it will persist the subscriptions to disk (or other persistent data storage medium), updating the absolute timeout based on the current time and a new timeout sent by the UCP (if it was different).
The NOTIFY Message
When a resource wishes to send an event notification, it will form a NOTIFY message of the following format:
The response is as follows::
This example of a GENA NOTIFY request and response demonstrates that a “upnp:propertychanged” event is being sent to http://danielwe/upnp:923. The USN header identifies “vcr.service1” as the event source. The XML contains the property name, value, and type. The “Seq” header indicates the sequence number of the notification. Sequence number 0 indicates the initial state update for the subscriber.
Property Change Event XML Schema
A UPnP property change event will be of the following form:
Here, a property named “foo” is of type “string” and has a value of “goodbye” and a property named “bar” has a type of “integer” and has a value of 27. The XML will be contains a list of multiple properties that have changed, along with a count to make it easy to determine this.
The UNSUBSCRIBE Message
When a UPnP UCP wishes to unsubscribe to event notifications for a UPnP service, it will form an UNSUBSCRIBE message of the following format:
The response would be as follows::
This example of a GENA UNSUBSCRIBE request and response demonstrates that the UCP is no longer interested in receiving event notifications from http://vcr.local/service1:200.
Step By Step: UCP to CD & Back
This section will take a step by step approach to what happens on both sides (UCP & CD) of an event notification. The description starts at the initialization of a UPnP device.
CD-Initiated NeedsSync Method
This method begins with the CD sending its initial state to the subscriber the first time an event is submitted by the service. UCPs will subscribe to the service first, then receive notifications for events as they occur. The first event will happen to be the initial state of the service. The UCP state table will always be in sync with this method.
When the CD sends a notification to a subscriber and receives an error. In this case, it marks the subscriber as “NeedsSync” and the next time an event is submitted, all events are sent to the subscriber. The problem with this is that the API needs to keep track of which subscribers need syncing and which ones don't. The client of this API (the UPnP service) would need to send separate messages to each subscriber and know which ones needed all events and which ones just wanted the ones that changed.
This method states that the UCP should subscribe to event notifications, then call a function that obtained the state from the service. This means that any events that were received in the meantime would need to be matched against the incoming set of events and replaced if they were older. This method leads to synchronization issues where the UCP may receive events that are newer but when it queries for the state, it gets an older view of the table. This requires using sequence numbers to determine which information is newer. If the view of the table received by the query is too old, it has to be discarded. Alternatively, the properties that were not received by event notification would not be overwritten, but all other properties would be. Using sequence numbers make this more complicated.
This preferred method takes a simpler approach. Any time the UCP subscribes to a service, the service will immediately afterwards, send the entire contents of the state table with the first notification. This precludes the UCP from making a query for the state table. Subsequent events update the local state table on the UCP. If the connection is lost, the UCP will lose its subscription. If the UCP realizes it has not received an event after a certain amount of time has elapsed, it will re-subscribe. At that point, the CD will re-send the entire state table again, and the UCP is ensured to be up to date.
Exemplary Computer Hardware
The system bus may be any of several types of bus structure including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of conventional bus architectures such as PCI, VESA, AGP, Microchannel, ISA and EISA, to name a few. The system memory includes read only memory (ROM) 824 and random access memory (RAM) 825. A basic input/output system (BIOS), containing the basic routines that help to transfer information between elements within the computer 820, such as during start-up, is stored in ROM 824.
The computer 820 further includes a hard disk drive 827, a magnetic disk drive 828, e.g., to read from or write to a removable disk 829, and an optical disk drive 830, e.g., for reading a CD-ROM disk 831 or to read from or write to other optical media. The hard disk drive 827, magnetic disk drive 828, and optical disk drive 830 are connected to the system bus 823 by a hard disk drive interface 832, a magnetic disk drive interface 833, and an optical drive interface 834, respectively. The drives and their associated computer-readable media provide nonvolatile storage of data, data structures, computer-executable instructions, etc. for the computer 820. Although the description of computer-readable media above refers to a hard disk, a removable magnetic disk and a CD, it should be appreciated by those skilled in the art that other types of media which are readable by a computer, such as magnetic cassettes, flash memory cards, digital video disks, Bernoulli cartridges, and the like, may also be used in the exemplary operating environment.
A number of program modules may be stored in the drives and RAM 825, including an operating system 835, one or more application programs 836, other program modules 837, and program data 838.
A user may enter commands and information into the computer 820 through a keyboard 840 and pointing device, such as a mouse 842. Other input devices (not shown) may include a microphone, joystick, game pad, satellite dish, scanner, or the like. These and other input devices are often connected to the processing unit 821 through a serial port interface 846 that is coupled to the system bus, but may be connected by other interfaces, such as a parallel port, game port or a universal serial bus (USB). A monitor 847 or other type of display device is also connected to the system bus 823 via an interface, such as a video adapter 848. In addition to the monitor, computers typically include other peripheral output devices (not shown), such as speakers and printers.
The computer 820 operates in a networked environment using logical connections to one or more remote computers, such as a remote computer 849. The remote computer 849 may be a server, a router, a peer device or other common network node, and typically includes many or all of the elements described relative to the computer 820, although only a memory storage device 850 has been illustrated in
When used in a LAN networking environment, the computer 820 is connected to the local network 851 through a network interface or adapter 853. When used in a WAN networking environment, the computer 820 typically includes a modem 854 or other means for establishing communications (e.g., via the LAN 851 and a gateway or proxy server 855) over the wide area network 852, such as the Internet. The modem 854, which may be internal or external, is connected to the system bus 823 via the serial port interface 846. In a networked environment, program modules depicted relative to the computer 820, or portions thereof, may be stored in the remote memory storage device. It will be appreciated that the network connections shown are exemplary and other means of establishing a communications link between the computers may be used.
In accordance with the practices of persons skilled in the art of computer programming, the present invention is described below with reference to acts and symbolic representations of operations that are performed by the computer 820, unless indicated otherwise. Such acts and operations are sometimes referred to as being computer-executed. It will be appreciated that the acts and symbolically represented operations include the manipulation by the processing unit 821 of electrical signals representing data bits which causes a resulting transformation or reduction of the electrical signal representation, and the maintenance of data bits at memory locations in the memory system (including the system memory 822, hard drive 827, floppy disks 829, and CD-ROM 831) to thereby reconfigure or otherwise alter the computer system's operation, as well as other processing of signals. The memory locations where data bits are maintained are physical locations that have particular electrical, magnetic, or optical properties corresponding to the data bits.
Exemplary Embedded Computing Device
With reference to
The device 900 also includes a network adapter 908 for connecting with a network media 910 that is interconnected with the computer network in which the authoritative names registry (described below) is implemented in accordance with the invention. The network adapter 908 can be a network interface card (or chip set integrated on a single board with the processing unit 902) appropriate to the particular network media 910. The network media can be any of various wired or wireless network media, including Ethernet, IEEE 1394 (a.k.a. firewire), radio frequency (including satellite, cell, pager, commercial signal sideband, etc.), power line carrier (PLC), phone line, and television cable, among others.
With reference now to
In non-configured (ad-hoc) networks, UPNP uses the AutoIP protocol.
AutoIP is an enhancement to DHCP that allows devices to claim IP addresses in the absence of a DHCP server or similar IP configuration authority. IP addresses are claimed from a reserved range that is not allowed to be transmitted on the open Internet; thus they are only good for the local network. The embedded computing device 100 claims an address by randomly generating an address in the reserved range and then making an ARP request to see if anyone else has already claimed that address. AutoIP systems will continually check for the presence of a DHCP server so that if one should ever come online, all the AutoIP devices will attempt to switch their IP addresses to one provided by the DHCP server. This allows a network to operate in isolation, be connected to the Internet with DHCP support and then to be returned to isolation. This type of scenario will be common in homes that use dial-up access.
The UPNP protocol also uses Multicast DNS for addressing the embedded computing device 900. The Internet Domain Name System (DNS) is a mapping system that translates human readable domain names, like microsoft.com, into their equivalent IP address. Most corporate intranets implement an internal version of the same technology to provide the same services. In small networks, such as at home or in small business, DNS servers may not exist. Multicast DNS allows DNS requests to be multicast. This allows a machine to see requests for its own name and respond to them. Like AutoIP, Multicast DNS is only used when a DNS server is not available. (For more information, see B. Woodcock, Zocolo, and B. Manning, “Multicast Discovery of DNS Services,” which can be found at http: search.ietf.org/internet-drafts/draft-manning-multicast-dns-01 txt.)
UPNP implements a peer discovery mechanism that uses the Simple Service Discovery Protocol (SSDP) for discovery of devices on IP networks. SSDP is based on profiles. A single identifier specifies a profile that defines a contract between the client and service (e.g., operational functions provided by the embedded computing device). By identifying itself with the profile, the service advertises compliance with the associated contract. Using a single identifier makes it possible to implement an extremely simple discovery system. Clients send out a User Datagram Protocol (UDP) multicast packet containing the identifier of the desired service on some standard channel. Services listen on the standard channel, read the request, see whether they provide the service, and respond if so.
UPNP also provides a Directories mechanism to allow discovery to scale—to the entire Internet if needed. When present, a directory will read all incoming service requests and respond to them itself. This requires that all services (e.g., the embedded computing device 900) register with the directory so that the directory is able to properly answer on their behalf. The directory is also responsible for communicating with other directories in order to determine whether the service is available within the local network, the WAN and potentially the Internet.
To simplify the discovery protocol, directories are treated as proxies. A proxy is a service that accepts requests and takes responsibility for finding the proper response. When a client comes online, it will perform discovery for the proxy. If the proxy is present, then the client will send all future discovery requests to the proxy. If the proxy isn't present, then the client will send all discovery requests to the reserved discovery multicast channel. Regardless of the presence of a proxy, the client's request format and procedures will always be the same. The only difference will be the address to which the client sends its requests. For services, the difference between a proxied and unproxied network is their need to answer discovery requests. On a proxied network, services need do nothing once they have registered with the proxy. On an unproxied network, they answer discovery requests directly.
SSDP uses the UDP- and Transmission Control Protocol (TCP)-based Hyptertext Transport Protocol (HTTP) to provide for service discovery. SSDP uses a Uniform Resource Identifier (URI) to represent the service and the OPTIONS method to provide for discovery. SSDP also will provide support for proxies. These proxies, which are really just fronts for directories, redirect discovery requests to themselves. It is the proxy's job to collect announce requests in order to determine what services are available as well as to communicate with other proxies in order to provide for scalable service discovery.
The discovery process returns only the basic information needed to connect to the embedded computing device. Once a service has discovered its peers, the service often needs to find out more information in order to work best with them. The description process returns a schema providing descriptive data about the service.
A schema is a structured data definition that defines a set of structured values that provide descriptive information about a service. UPNP uses the Extensible Markup Language (XML) for schema, because XML's self-describing structured data format provides the level of expressiveness and extensibility needed by a universal schema and data format.
Accordingly, UPNP supports automatic network introduction, meaning that devices and their related services have the ability to be self-describing and allow automatic configuration. When a device is plugged into the computer network, the device automatically configures itself and acquires a TCP/IP address. The device then announces its presence to other devices already on the network using a simple discovery protocol based on the Internet HTTP protocol and is immediately ready to share its services with any device that requests them.
With UPNP, device developers are not required to develop specific device drivers to operate under UPNP. The task of preparing a device for operation in this network environment thus is fairly simple. Moreover, in configured networks, dynamic detection allows an operating system to immediately begin using added devices or stop using removed devices without rebooting.
UPNP Devices support automatic discovery, identification, and configuration to achieve interoperability in the home environment, but must also operate correctly in a managed corporate network. Devices can be networked instead of being attached directly to a PC, and devices are all autonomous citizens on the network, able to talk with each other and exchange information. UPNP provides a unified way of performing directory services with automatic configuration. Capability for simple discovery mechanism used in the home environment provides the ability for any device to become a node on the global Internet. Additionally, directory services can be leveraged if they are available in the corporate environment.
UPNP provides a common set of interfaces for accessing devices and services, enabling the operational unification of diverse media types. Communications protocols for Universal Plug and Play are based on industry standards, especially key Internet standards such as TCP/IP, HTML, XML, HTTP, DNS, LDAP, and others. Individual implementations for particular networks and buses are built on established protocols.
As shown in
The simple discovery 924 is a software code module (about 4 Kbytes) that implements a simple discovery procedure (described below) for automatic network introduction under the UPNP protocol.
The simple discovery procedure additionally provides an Extensible Markup Language (XML) format device description, which is downloaded to clients that access the device to allow activation of device functionality from the client. XML is a textual, tag-based markup language. It was originally designed to be the “webby” simplification of SGML (Standard Generalized Markup Language), and is therefore intended to be used to create “vocabularies” of tags that can be used to apply semantic markup to documents, such as who the author was, what constitutes a paragraph (semantically, not from a display point of view), when the author last had breakfast, and so on. (For more information, see A. Layman, E. Jung, E. Maler, H. Thompson, J. Paoli, J. Tigue, N. H. Mikula, S. De Rose, “XML-Data”, which can be found at http://www.w3.org/TR/1998/NOTE-xml-data-0105; and MSDN Online, XML Data Center at http://msdn.microsoft.com/xml/default.asp.) In the context of UPNP, XML is used to provide the description of services and capabilities of the embedded computing device. The embedded computing device makes its features visible to clients by providing its XML device description, which the client can use to activate device functions 922. For example, if the device is a camera, the client's browser can direct the camera to zoom in/out or adjust contrast using the mechanism of XML.
The XML device description can provide links (via a uniform resource locator or URL address) to an accompanying XSL format style sheet. The XSL style sheets are used to present the data in different ways, i.e., the style sheets are applied to present different views of the same data. For example, if the device contains a file system, one style sheet can show the file selections; another shows the file sizes in some sort of diagram; yet another style sheet could make thumbnails of these image files.
The HTTP 925 is a software code modules (about 20 Kbytes) that implements the standard HTTP protocol, which is an open standard mechanism for client/server message-based communication. HTTP provides for proxying, content negotiation and security. (For more information, see R. Fielding, J. Gettys, J. Mogul, H. Frystyk, T. Berners-Lee, “Hypertext Transfer Protocol—HTTP/1.1”, which can be found at http://www.ietf.org/rfc/rfc2068.txt.)
The TCP/IP stack 926 implements the standard TCP/IP networking protocols for communication on the computer network. The Internet Protocol (IP) is the foundation protocol of the Internet. It defines how a single message is sent from a source through zero or more routers to its final destination. It covers issues such as message length, message fragmentation, addressing, and routing concerns. The Transmission Control Protocol (TCP) is an IP-based protocol that provides support for the reliable, ordered delivery of messages over IP. Additionally, User Datagram Protocol (UDP) and Internet Group Management Protocol (IGMP) multicast send/listen capability are included in the implementation.
The Autonet 928 is a software code module also used for automatic network introduction via AutoIP in the UPNP protocol. Autonet uses a predefined set of IP addresses and, when a device is connected to the network, it pings an address in this address space. If it gets no replies, the device assumes that the address is available and assigns it to itself. To make this functionality even more useful it is combined with Multicast DNS, in which the device itself holds its own name. Thus it is not even necessary to determine what IP address the device assigned to itself, because its name can always be used instead. An IP Multicast is a mechanism for sending a single message to multiple recipients. IP multicasting is especially useful for discovery operations where one does not know exactly who has the information one seeks. In such cases, one can send a request to a reserved IP multicast address. Any services that can provide the requested information will also subscribe to the multicast request and thus be able to hear the information request and properly respond. Multicast DNS is a proposal to the IETF on rules for making normal DNS requests using multicast UDP. (For more information, see B. Woodcock, B. Manning, “Multicast Discovery of DNS Services”, which can be found at http://www.ietf.org/internet-drafts/draft-manning-multicast-dns-01.txt.)
The DHCP 930 is a software code module that implements the Dynamic Host Configuration Protocol (DHCP), which is a mechanism for providing devices with configuration information needed to access the Internet. The mechanism functions through the use of a multicast request for configuration information that is generally responded to with an IP address and DNS server location. Additional information can only be returned in the response.
At the announce phase, the embedded computing device 900 sends out a small multicast packet so that other devices can find it on the network. The multicast message packet essentially says, “I am here, I am, (say), a camera, and you can reach me at this IP address or URL.”
At the discovery phase, the embedded computing device 900 listens for a discovery packet coming from a simple discovery client, i.e., the device announces itself, then listens for discovery. The discovery packet also is sent out by multicast.
At response to discovery, the embedded computing device 900 listens to the multicast address and then parses the information from a Simple Discovery request to decide if the request is for its kind of device. If so, the device 100 then sends back a response packet containing the following information: the IP address or URL where it can be reached; identification of its own device type; and the discovery packet ID so the requesting client knows which request is being answered.
At the Autonet phase, the Autonet module 928 of the embedded computing device 900 uses a predefined set of IP addresses and, when the device is connected to the network, it pings an address in this address space. If no reply is received, the device 900 assumes that the address is available and assigns it to itself. Alternatively, the device 900 may combine Autonet with Multicast DNS, and itself hold its own name. In which case, it is not necessary to determine what IP address the device assigned to itself, because its name can always be used instead.
Both the Announce and Discovery packets also contain a link or a URL to an XML file that is used by the embedded computing device at the device description phase to describe itself (i.e., its functionality). This XML data contains all the facts about the device. XML can also have URLs that point to appropriate style sheets (XSL files) that are used for optimal presentation. The XSL style sheets are used to present the data in different ways, i.e., the style sheets are applied to present different views of the same data. For example, if the device contains a file system, one style sheet can show the file selections; another shows the file sizes in some sort of diagram; yet another style sheet could make thumbnails of these image files.
With reference now to
The simple discovery 954 is a module that implements the above-described simple discovery per the UPNP protocol. The XML 955 is a module that processes the XML device description and XSL style sheets for presentation in the application's user interface. The LDAP 956 implements the standard LDAP directory protocol for name look-up. The TCP/IP stack 958 implements the TCP/IP protocol for communications over the computer network.
Illustrative Pervasive Computing Environment
Various embedded computing devices also connect to the computer network via various network connections to the PCs 1002, 1004. These include an audio device 1014 (e.g., speakers, radio tuner, microphone), and printer 1015 which connect to the PC 1004 through a USB 1017. Also, a digital camera 1020, a handheld PC (H/PC) 1021 and another personal computing device 1022 connect via an infrared port (IRDA) 1024, which also attaches to the PC 1004 through the USB 1017. Also, lighting switches 1030 and like home appliances are connected via an A/C power line-based networking 1032 to the PC 1002. Further, a chain of IEEE 1394 cables 1048 connect a digital TV 1040, DVD player 1041, digital video camcorder (DV/DVC) 1042, an audio device 1043 (e.g., CD player/recorder, radio receiver, amplifier, and like audio system component), and a game console 1044. Devices, such as a portable telephone 1050 and remote control 1051, have a radio frequency network connection with the PC 1004.
With their various inter-networked connections, the embedded computing devices are “visible” and accessible from a client device 950 (
Contract Definition Language
Contracts describe the public behavior of UPnP devices, and alternatively of other entities on the Web (reachable via HTTP, mostly) or other computer network that react to and emit messages. The Contract is written in a Contract Definition Language (CDL). The messages for the most part are structured documents, e.g., in XML. The messages may also be HTML pages, streaming media, images or other datatypes appropriate to the WebObject.
The contract will describe the following attributes of a WebObject:
All of these attributes may not be present in the contract as some of them (the end-point, for instance) may not be available at development time.
WebObjects can be accessed using multiple protocols: HTTP, GENA, SMTP, FTP, MSMQ, . . . This section discusses how to describe the protocol bindings particular to a WebObject. The templates for describing the protocol use the format:
The “protocol” element may have an “id” attribute. This is useful when multiple messaging patterns will use the same protocol definition. This will be covered in more detail below.
For the sake of convenience, we only cover HTTP-based protocols here. Extending this model to cover the other protocols is straightforward.
GET with query string
This description indicates that the following are valid URLs:
The reason for not associating the query variables with the GET verb is because it is valid to send a POST message to a URL containing query variables.
The “value” attribute for the “QUERY” element implies that the value is static—it is to be treated as a part of the URL. Declaring it this way allows the appropriate construction of the query string to be handled by the caller.
The default attribute indicates that the parameter's value can be changed.
The M-POST and the enclosed MAN elements declare the mandatory extension mechanism to be used. The optional extension mechanism can also be handled in this way.
The “HEADER” element allows the declaration of HTTP headers to be used.
Below is an example of an XML payload description.
Using this declaration, the below are valid XML fragments:
The messaging pattern declaration acts as an anchor for pulling together the protocol, delivery characteristics and the payload information. The messaging pattern declarations can include these types.
Request/response (RR). The RR pattern is named. The two samples below are equivalent mechanisms for declaring the protocol to be used for the RR messaging pattern. The linking mechanism is useful when multiple RR pairs use the same protocol data. This is the case for UPnP. Also, a service may employ multiple protocols for achieving the same “method”-call. The “is” attribute accepts a list of ID-Refs—implying that either of the protocols are equally suitable for accessing the functionality.
The payloads for request, response and error, in case of XML data, are identified by the names of the elements referenced by the “is” attribute. The schema information is assumed to be in the same document. Below are examples using the two schemes:
The CDL described herein keeps the element declarations in the “schema” block rather than sprinkle them in with the messaging pattern definitions. The reasons for this are:
In case the request or response are not XML documents but HTML documents, or binary files, the following syntax will be used. The contained element defines the nature of the data. The use of MIME is not in the HTTP-specific sense but in the “nature of the payload” sense. The presence of the “is” attributes indicates that the MIME type is “text/xml.”
The contract may specify the delivery characteristics (sometimes also referred to as quality of service) required or supported by the server. Examples are:
Having described and illustrated the principles of our invention with reference to an illustrated embodiment, it will be recognized that the illustrated embodiment can be modified in arrangement and detail without departing from such principles. It should be understood that the programs, processes, or methods described herein are not related or limited to any particular type of computer apparatus, unless indicated otherwise. Various types of general purpose or specialized computer apparatus may be used with or perform operations in accordance with the teachings described herein. Elements of the illustrated embodiment shown in software may be implemented in hardware and vice versa.
In view of the many possible embodiments to which the principles of our invention may be applied, it should be recognized that the detailed embodiments are illustrative only and should not be taken as limiting the scope of our invention. Rather, we claim as our invention all such embodiments as may come within the scope and spirit of the following claims and equivalents thereto.