CA2265478A1

CA2265478A1 - Method and apparatus for optimizing a medium access control protocol

Info

Publication number: CA2265478A1
Application number: CA002265478A
Authority: CA
Inventors: Hilton K. Hong; Juan Grau Jr.; Arthur B. Coleman
Original assignee: Individual
Current assignee: Proxim Wireless Corp
Priority date: 1996-09-23
Filing date: 1997-09-23
Publication date: 1998-03-26
Also published as: EP0909493A1; CN1232587A; JP2001501053A; US5844900A; KR20000029996A; WO1998012829A1; AU4583797A; BR9711531A; AU733369B2

Abstract

The present invention is directed to providing a wireless medium access control protocol wherein throughput can be optimized as the communication system dynamically changes when nodes (102, 104, 106, 108, 110, 112, 114, 116) either join or leave the communication system (100).

Description

?W0 98/1282910152025CA 02265478 1999-03-03PCTIU S97! 16356METHOD AND APPARATUS FOR OPTIMIZING AMEDIUM ACCESS CONTROL PROTOCOLBACKGROUND OF THE INVENTIONField of the Invention: The present invention relates generally to communication among pluralnodes, and more particularly, to control of wireless communications byadjusting a number of deferral and fairness slots.State of the Art:Communication systems, such as local area networks (LANS), haveachieved widespread use for the exchange of information. These systemsinclude a plurality of data processing nodes, or "agents", which contend foraccess to a communication medium that interconnects the nodes.For example, commonly assigned U.S. Patent No. 5,231,634 (Giles etal), the disclosure of which is hereby incorporated by reference in its entirety,relates to a communications system wherein a predetermined number of deferraland fairness slots are used. In the communication system of the Giles patent, aplurality of nodes are used, each of which is equipped with a radio transceiver.The â634 patent describes medium access control of a wireless LAN to providedecentralized, peer-to-peer access for all of the agents. A peer-to-peercommunication system is one in which all agents can freely communicatewithout any agent having an advantage over another.Wireless communication systems require consideration of potentialproblems which do not exist in hardwired systems. For example, in a wirelesscommunication system, the nodes can be mobile, moving in and out of rangewith other nodes in the system. The â634 patent is directed to continuouslyensuring that all agents possess fair access to the communication system (i.e.,to avoid, for example, agents which are physically closer to a given destinationfrom having an unfair advantage due to their signal strength being greater thanthat of agents physically located further from the destination).?WO 98/128291015202530CA 02265478 1999-03-03PCTlUS97I16356When using a wireless medium access protocol such as that described inthe â634 patent, the number of deferral and fairness slots can significantly affectthe amount of throughput in the system. In a system where a large number ofwireless nodes are contending for medium access at the same time, a largernumber of fairness and deferral slots is desired. In contrast, when only one ortwo wireless nodes are contending for medium access, fewer slots are desired.Thus, when a system is configured with a predetermined number of slots, thesystem may be optimized for operation with a given number of active nodes,yet be inefficient when that number of nodes is changed.Accordingly, it would be desirable to provide a wireless medium accesscontrol protocol, which can be optimized for any number of nodes currentlyactive in the communication system.SUMMARY OF THE INVENTIONThe present invention is directed to providing a wireless medium accesscontrol protocol wherein throughput can be optimized as the communicationsystem dynamically changes when nodes either join or leave the communicationsystem.Generally speaking, exemplary embodiments of the present inventionrelate to an apparatus and method for controlling wireless communicationamong plural nodes in a communication system comprising: means formonitoring a number of transmissions which occur within the communicationsystem; and means for dynamically changing a number of slots for which nodesof the communication system contend for access to a wireless communicationmedium of the communication system, said step of dynamically changing beingperformed as a function of said number of monitored transmissions.BRIEF DESCRIPTION OF THE DRAWINGSThe present invention can be further understood with reference to thefollowing description and the appended drawings, wherein like elements areprovided with the same reference numerals. In the drawings:Figure 1 illustrates a multiple masterânode configuration for anexemplary communication system in accordance with the present invention;-2-?W0 98/128291015202530CA 02265478 1999-03-03PCT/US97/16356Figure 2 shows a frequency hop period in accordance with an exemplaryembodiment of the present invention;Figure 3 shows an arrangement of a variable number of slots in acontention period with fairness; andFigure 4 shows a ?owchart of operation in accordance with anexemplary embodiment of the invention.DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS1. OverviewFigure 1 shows an exemplary embodiment of a communication system100 for controlling communication of frequency hopped signals among aplurality of nodes 102-116 (e.g., communication stations). It should beunderstood that only nodes which are synchronized to one another cancommunicate with one another. A frequency hopping medium access protocolcan be used to control communication among the nodes in server basednetworks, peer-to-peer networks, ad hoc networks and so forth.In Figure 1, only nodes which are synchronized to one another cancommunicate with one another. A frequency hopping medium access protocolin accordance with the aforementioned copending applications can be used tocontrol communication among the nodes in server based networks, peer-to-peernetworks, ad hoc networks and so forth.Frequency hopping within a communication system is performed bytransmitting/receiving information using any number N of different frequencies(e.g., a prime number, such as 79 or 83 different frequencies). Thecommunication system repeatedly switches frequencies during communication,sequencing through all N frequencies of the hop sequence. The frequencies areswitched in either a random or a predetermined sequence (i.e., the hopsequence). A random sequence can be used if a mechanism is in place wherebyboth a transmitting node and a receiving node use the same random generationtechnique, or if the transmitter communicates the random sequence to thereceiver.. ,.,.e..................................r.......................- . . ......., .?W0 98/ 128291015202530CA 02265478 1999-03-03PCTlUS97l16356The system remains at each of the different frequencies of the hopsequence for a selected period of time, referred to herein as the hop period. Inaccordance with exemplary embodiments, the hop period can be fixed at anyone of a plurality of different time lengths. For example, in an exemplaryembodiment the hop period can be selected to be 100 milliseconds (ms), 200 msor 400 ms. The hop period is selected by the master node which is in controlof synchronization of the communication system at a given time. Once the hopperiod is set by the master node, it will not vary unless the master nodechanges the hop period or the communication system resynchronizes to a newmaster node with a different hop period. All nodes which are synchronized to aparticular master node recognize the hop period defined by that master node.Communication among the various nodes of the communication systemcan be achieved by synchronizing these nodes to the same frequency within thehop sequence, and by aligning the hop periods so that all nodes incommunication with one another switch frequencies together. In accordancewith exemplary embodiments, synchronization is maintained continuously, andnot only when a node attempts to initiate or receive communications. Thus,data and other information can be transmitted whenever the communicationsystem is open (i.e., contention-free), without having to wait for asynchronization period to elapse.A decentralized or distributed method of synchronization can be usedwhereby any one of multiple nodes can control system synchronization andcommunication. Accordingly, each of nodes 102-116 in Figure 1 is assigned ahierarchical designation representing a priority of the node to controlsynchronization of the communication system. The node which controlssynchronization at a given time will be referred to hereafter as a "master node".Any remaining nodes which are in communication range of the master node canbe synchronized to the master node. For example, node 102 of Figure 1 can beconsidered a master node for controlling synchronization of the communicationsystem. However, if node 102 is disconnected (e. g., a mobile communication?W0 98/128291015202530CA 02265478 1999-03-03PCT/U S97/ 16356station which leaves the building), the remaining nodes resolve which node willbecome the new master node.In Figure 1, each node is assigned a hierarchical designationrepresenting a priority of the node in contending to become a master node. Inan exemplary embodiment, the hierarchical designations are of three differenttypes: (1) a designation of highest priority referred to herein as master stationpriority; (2) an intermediate priority designation referred to herein as analternate master station priority which is of a priority lower than that of amaster station priority; and (3) a lowest priority designation representing nodeswhich can never control synchronization.In the above example, the "masterânode" can be a node of either masterstation priority or a node of alternate master station priority. Nodes of priority(3) cannot communicate within a communication system unless a master node islocated within their range of communication to control synchronization. Thenodes of the communication system resolve which node will be the master nodefor controlling synchronization within the system at any given time.As referenced herein, the phrase "communication system" refers to oneor more nodes which can communicate with each other. Because the nodes aremobile, one or more dynamically reconfigured sub-systems may exist within thecommunication system. For example, a first plurality of nodes may be withincommunication range of one another and controlled by a first master node. Asecond plurality of nodes within the communication system may be incommunication with one another and controlled by a second master node.Depending on the priority assigned to each node, it is possible that a singlenode which can communicate with some of the first plurality of nodes and withsome of the second plurality of nodes will attempt to become a master node.Whether the single node will be successful or not depends on master noderesolution.2. Communication Formata. Hop Period ?WO 98/128291015202530CA 02265478 1999-03-03PCT/U S97] 16356Figure 2 illustrates a time oriented relationship of an exemplary hopperiod 200 with the total time duration selected for the hop period beingindicated in a synchronization message 204 (i.e., Sync MSG) from the masternode. The phrase "hop period" can be abbreviated to "hop".A master node controls synchronization of nodes in the communicationsystem by transmitting the sync message. A beacon period is included in thehop period as an aid to facilitate synchronization. The beacon period permitsfaster synchronization of a node to a master node. A node receiving the beaconcan lock onto the frequency of the beaconing master node prior to receipt of async message. Each of the sections set forth in the exemplary Figure 2 hopwill now be described.(1) Beacon Period â The beacon period ensures that there is energypresent at the current frequency (i.e., a beacon frequency) long enough for anyreceiver node to scan all possible frequencies in the hopping sequence. Inaddition, information is included in the beacon to allow a receiver, once it hasdetected energy in a particular frequency, to determine whether this beacon hasthe appropriate channel/subchannel information. To detect a beacon, nodescontinuously sequence through all N frequencies of the frequency hoppingsequence looking for a beacon transmitted by a master node. The master nodesends the beacon for a long enough period of time to enable a node to scan allN frequencies during a scanning period. The length of time the master nodesends the beacon to ensure that other nodes will have an opportunity to scan allN frequencies is the beacon period 202.The beacon period 202 is a fixed duration, contention-free period oftime during which a beacon message is broadcast by the master node. Thisfeature ensures that a node of master station priority can immediately transmitinformation for establishing communication with other nodes of thecommunication system, without first monitoring the communication mediumbefore transmitting such information. This avoids delays typically required fora node to transmit such information in, for example, a server-basedcommunication system. The master node can be configured to beacon, for-6-?WO 98/128291015202530CA 02265478 1999-03-03PCTIUS97/16356example, every 1 to 8 hops for the purpose of optimizing sync times versusoverhead for a given installation.One example of a beacon is a signal which includes the followingsequence: a first delay period (i.e., Delay 1); a Preamble; a second delayperiod (i.e., Delay 2); and a beacon message. The four components of thissequence can be repeated any number of times, endâto-end, for an entire lengthof beacon period 202. Each of these four components can be described asfollows:Delay 1 This delay is the time it takes to change from switching tosend a Beacon Message to the time it takes to send aPreamble message.Preamble The inclusion of a Preamble makes it easier for nodes tolock onto a particular master node (e.g., a waveformconstructed to provide easy lock, which is appended to thestart of a data packet).Delay 2 This is the time it takes to transition from sending aPreamble to sending a Beacon Message.Beacon Msg. A beacon message corresponds to a short data packet for tellingother nodes (a) a channel at which the master node is operating;(b) a subchannel on which the master node is operating; and (c)at which frequency the master node is operating (i.e., which ofthe N frequencies). The beacon message is optimized forefficient searching by the remaining nodes. The Beacon Messagecan include, for example, 5 data bytes, 2 ?ag bytes, and 2 cyclicredundancy check (CRC) or checksum, bytes. Flag bytesrepresent framing bytes in between which data can betransmitted.The foregoing beacon sequence is by way of example only, and those skilled inthe art will appreciate that any signal or sequence of signals can be used as thebeacon. For example, the above beacon could be modified to exclude thepreamble and/or the delays. The beacon could merely be an energy bearingsignal for indicating to receivers that a sync message will follow.An appropriate beacon period should be selected to account for any timea receiving node requires to prepare for reception of each frequency of the-7-?W0 98I128291015202530CA 02265478 1999-03-03PCT/U S97/ 16356frequency hopping sequence (i.e., the radio period, or key-up time) and thetime the receiver must listen at a frequency of the frequency hopping sequenceto detect the beacon message (i.e., beacon listening time).The radio key-up time can, in an exemplary embodiment, include thetime required to receive a preamble within a beacon period. A worst casesituation occurs when the preamble has just been missed by a receiving nodesuch that the node must wait for a subsequent preamble to be sent. Thus, aminimum radio key-up time should correspond to a sum of two preambles,delay 2, a beacon message and delay 1.A beacon listening time corresponds to an amount of time required for anode to listen at the frequency at which the preamble was detected during ascan through all N frequencies. The beacon listening time must be long enoughto guarantee that a beacon message sent at that frequency will be detected. Aworst case occurs when a node fails to detect a beacon message such that areceiver of the node must wait for a subsequent beacon message. Therefore, inan exemplary embodiment, a beacon listening time corresponds to a sum of twobeacon messages, delay 1, delay 2 and l preamble.A beacon period can be calculated as the number N of availablefrequencies in the frequency hopping sequence multiplied by the radio period,plus the beacon listening time. Thus, a beacon period represents the timerequired to switch through each of the N frequencies, prepare the radio, andlisten at the beacon frequency long enough to detect a beacon message.In summary, the beacon message broadcast during the contentionâfreebeacon period permits other nodes within the communication system todetermine that they have identified a correct frequency of the hopping sequence,as well as the channel and subâchannel with which the master node isbroadcasting. However, the nodes controlled by the master node also need todetermine when to switch frequencies (i.e., synchronize their hop periods),since frequency switching must occur for all stations in the communicationsystem at the same time. Synchronization information for synchronizing to ahop period of the master node is not included in the beacon message. Rather,-3-?W0 98/128291015202530CA 02265478 1999-03-03PCT /US97l16356information pertinent to synchronizing is included in a sync message (i.e., SyncMSG).(2) Sync MSG - Sync message 204 enhances the ability of a node toquickly synchronize to a communication system. Although a beacon period isonly included in hop periods which have been designated to include a beacon(e.g., once every 8 hops), the sync message is included at the beginning ofevery hop period, regardless of whether a beacon message is broadcast withinthe hop period. A master node may only broadcast its contention-free beacononce every x (e.g., 8) hop periods to minimize its contention-free control of thecommunication medium. However, Sync message 204 contains pertinentinformation to allow nodes to sync to the master node, and this information canbe broadcast on every hop so that nodes can update their hop interval times andadjust for drift. The sync message can also contain pertinent informationconcerning a contention-free outbound period during which a master node canbroadcast information to all nodes of the communication system withoutcontention.When a beacon message has been located, nodes can identify whichchannel and frequency (i.e., hop number) the master node is currently on to aidin the synchronization process. The sync message is used to transmit a hoptimer count which re?ects a count remaining on a hop timer of the master nodewhen it sets up the sync message. The hop timer tracks the time duration of ahop period. Based on the remaining count transmitted by the master node inthe sync message, a node can subtract the appropriate count for processing andtransmission delays, and then write the remainder into a hop timer register sothat its hop timer interrupt will occur at approximately the same time that a hoptimer interrupt occurs in the master node. This allows the node to synchronizeits hop period to the hop period of the master node.(3) OutBound â A master node can communicate messages to allnodes with which it is synchronized in a communication system during acontention-free outbound period 206. The outbound period 206 can be used bymaster nodes to off load any queued data without contention. This allows-9- ?W0 98/ 128291015202530CA 02265478 1999-03-03PCT/US97ll6356optimum transfer of data for the master node, and can be used forconfigurations where the master node is, for example, also the server in aLAN. The outbound period is of variable length as required. For example,this period can be expanded in hops which do not include a beacon period. Amaster node can be configured to transmit a maximum number of messages thatcan be transferred during this period, or any number of messages less than themaximum, with the minimum equal to O.(4) Time Bounded â An optional, contention-free, time boundedperiod 208 can support time bounded services.The period of time represented by the beacon period, the sync messageperiod, the outbound period and the optional time bounded period collectivelyrepresents a first period of time, referred to herein as a contention-free "A"period. The remaining portion of the hop period is a contention period duringwhich nodes synchronized to one another within a communication system viefor access to the communication medium. This contention period is referred toherein as a "B" period which is initiated by what is referred to herein as a"start of B". During the contention period, nodes can contend for access to thecommunication medium using a contention protocol such as that described incommonly assigned U.S. Patent No, 5,231,634, the disclosure of which ishereby incorporated by reference in its entirety.(5) Start of B - A start of B message 210 is broadcast by the masternode to announce the start of a "B period" to all nodes.(6) B Period - A "B" Period 212 is a random access contentionperiod during which the master node and all other nodes can communicate withone another.Having discussed the contents of a hop period used by nodes tocommunicate in accordance with exemplary embodiments of the presentinvention, attention will now be directed to the scanning period. A scanningperiod corresponds to a period during which a node scans for a beacon tosynchronize to a master node. A scanning period is also used to search for-10-?W0 98/128291015202530CA 02265478 1999-03-03PCT/US97/16356beacons to detect the existence of other nodes (i.e., of master station priority oralternate master station priority) transmitting beacons.b. Scanning PeriodA scanning period corresponds to the time a nonâmaster node searchesbefore determining that a master node does not exist. For example, when amaster node sends a beacon at least once every 8 hops, nodes must locate abeacon within that period. In a normal hopping sequence, a transition fromone frequency to a next frequency occurs in a predetermined non-sequentialorder in accordance with exemplary embodiments. However, when a node issearching for a beacon, it will begin scanning with the current frequency it ison and go up in frequency in sequential order until the Nth frequency isscanned. The hopping sequence then wraps around to the first frequency andcontinues through the sequence of frequencies again.c. Frequency DriftingHop timers included in the nodes are resynchronized every hop period tocorrect for errors associated with tolerances of the timing references for eachnode. Each hop period, synchronized counters may be off by several counts,and if nodes are not resynchronized often, they can drift. Because the nodesare resynchronized every hop period-, frequency drifting is corrected in eachhop period.However, exemplary embodiments can also account for frequency driftat the beginning of each hop period. For example, the master node can delaysending the sync message for a predetermined period of time referred to hereinas a drift period. The master node need not delay sending the beacon duringhops which include a beacon period because it is not important for synchronizednodes to receive beacon messages. With the exception of the beacon, no data issent at the beginning or end of a hop period within the drift period.The drift period differs depending on the hop period length. Inexemplary embodiments, sync messages can be missed for a maximum of 4 hopperiods before a determination is made that synchronization has been lost. Inthis case, the error rate must be multiplied by 4. The result is multiplied by 2-11-?WO 981128291015202530CA 02265478 1999-03-03PCT/U S97/ 16356to account for the master node being off by a gain of 100 parts per million(PPM) while the node is off by a loss of 100 PPMs, creating an error rate ofapproximately .08%.(1. Message FormatTo provide communication between nodes using frequency hopping andsynchronization, a protocol for transmitting data and other information can beestablished in accordance with the aforementioned U.S. Patent No. 5,231,634.In accordance with exemplary embodiments, data can be transmitted in frames,with different frame types being employed as necessary. Each frame caninclude a frame header, and additional remaining unused bits which arereserved and set to zero unless otherwise specified.In accordance with exemplary embodiments, a standard frame headercan be used on all frames with the exception of a beacon message. A standardframe header includes a ?ag to indicate the start of the frame, an identificationcode of a transmitting node (e.g., whether the node is a master station, alternatemaster station, or a low priority station) control information, (e.g., transmissionpower level, data compression technique, radio speed, and so forth), mediumaccess control length (e.g., frame length from node/frarne up to CRC), securityinformation, and channel/subchannel. The node/frame type portion of thestandard header can specify antenna diversity (i.e., indicating which antenna isto be used as the sending node), the node type (i.e., master station, alternatestation or station) and the frame type. Different types of frames include abeacon frame, a resync frame, a resync acknowledgement frame, a sync frame,a requestâto-send frame (RTS), a clear-to~send frame (CTS), a request mediaaccess frame (RMA), a medium access control acknowledgement frame(MACK), a logical link control acknowledgement frame (LACK), a datamessage frame, a wakeâup message frame, and so forth.Source and destination addresses for information included in the framecan be included in an extended header definition specified by the user. Thestandard frame header can include a header followed by a destination addressand a source address.-12-?W0 98/ 128291015202530CA 02265478 1999-03-03PCTIU S97/ 16356A start of B message can be sent by a current master node to indicatethe start of a contention period. Nodes must hear the message to participate inthe contention period, and nodes, upon receiving this message, will use a slotdefinition and apply it to the current hop. The start of B message contains afield which defines the maximum time to wait for successive start of Bmessages on successive hops before starting contention without it.A beacon message can be energy (e. g., a message) which is sent at thebeginning of some hops by the master node. In an exemplary embodiment, thebeacon message can be a relatively short message that is repeated several times.A beacon message can include a ?ag to indicate the start of the beaconmessage, a frame type indicator to confirm that a beacon message is beingtransmitted, a security identification of the master node sending the beaconmessage, the channel and subchannel of the master, a CRC portion and anending ?ag.A resync message can, in accordance with exemplary embodiments,have one of two versions. A first version is a broadcast message sent by amaster in an outbound period to inform all nodes synchronized to the masternode that a second master node has been located to which all other nodesshould synchronize. This first version of the resync message requires noacknowledgement. A second version of the resync message can be a directedmessage from a node designated as an alternate master to a first master stationnode during the contention period to inform the first master station that asecond master station node exists. The alternate master node sends informationabout the second master to the first master station node. Upon receipt of thisinformation, the first master station sends a resync acknowledgement messageto inform all nodes of the identity of the new master (i.e., the identity of eitherthe first master node or the second master node, depending on a masterresolution protocol). This message includes the medium access controlextended information, the channel/subchannel of the new master, the hopnumber which serves as an index into the channel frequency table to find a syncmessage, a hop duration, CRC information and an ending ?ag. A-13-?WO 981128291015202530CA 02265478 1999-03-03PCTIUS97/16356resync acknowledgement message is sent from a master node and broadcast toall stations. This acknowledgement message can be sent in response to a nodesending a resync message to the master node, and it informs all nodessynchronized to the master node to resynchronize to a new master at thefrequency specified in the message. Thus, a resync acknowledgement messageincludes an extended medium access control, a channel/subchannel of the newmaster, a sync hop number used to locate the sync message, a hop duration, anew master identification to which all communications should synchronize, aCRC portion and an ending ?ag.The Sync message sent by a master node at the beginning of each hopperiod contains information for nodes to adjust their hop timer counters tomaintain synchronization from hop to hop. A synchronization message includesan extended medium access control portion, a node name (e.g., ASCII) of themaster node, a control portion for defining the hop (i.e., whether next hopwill/will not contain beacon, whether hop is dedicated to searching, or whetherregular hop), a hop number, a hop duration, a CRC portion and an ending ?ag.3. Communication System Operationa. InitializationIn accordance with the present invention, exemplary embodimentsinclude a step of initializing the communication system by synchronizing thenodes to a first node. Initialization of a node designated with a master stationpriority is relatively straightforward. When a node designated with a masterstation priority is requested to initialize, it immediately sets its hop timer andstarts beaconing and sending sync messages. Because it is of the highestpriority, it does not search for other nodes of master station priority beforeattempting to assume a role as a master node itself. Nonâmaster nodes(e.g., a low priority node or a node of alternate master station priority which isnot currently operating as a master node) initialize by first scanning for abeacon during a scanning period. Depending on a configuration of thecommunication system, a node will scan for a beacon which matches a subset-14-?W0 98Il28291015202530CA 02265478 1999-03-03PCT/U S97] 16356of its channel and subchannel (as well as any security identificationinformation) .Because the detail regarding initialization of synchronization in acommunication system, the searching for multiple master stations, the searchingfor multiple nodes (alternate stations) and the synchronization to a newlydetected master station do not in and of themselves, constitute a portion of thepresent invention, and because such features are described in theaforementioned copending applications, further discussion of these features isunnecessary.b. Medium Access Control OptimizationIn a medium access control protocol as described in the â634 patent,access to a communications medium shared by multiple nodes is controlled bysensing a communication medium by a first node to determine if thecommunications medium is in use. If not, the first node transmits a requestâto-send message that includes reservation duration information. The request-to-send message is received by a second node to which information is to be sent.Upon receiving the requestâto-send message, the second node transmits a clear-toâsend message which includes reservation duration information on behalf ofthe first node, after which the first node then transmits information to thesecond node.A node which is transmitting data first calculates the amount of time(i.e., reservation time) the entire transmission process will take, and thenselects a random "slot" time to begin the transmission. If the communicationmedium is clear, the sending node will transmit an RTS message, and thenawait receipt of a CTS frame from the receiving node. All nodes that receiveeither an RTS or a CTS will defer from contending for the time specified ineither frame (i.e., both the RTS and CTS frames contain the reservation time)during typical operation.As described in the â634 patent, to preserve fair access to the network,the contention time is broken into a fixed number of slots for which the nodescontend to obtain access to the wireless communication medium. For example,referring to Figure 3 of the present application, the contention time is broken-15-?WO 981128291015202530CA 02265478 1999-03-03PCT/U S97/ 16356into two fixed periods: a fairness period 302 and a deferral period 304, both of 9which are divided into a fixed number of slots. The fairness period precedesthe deferral period. Nodes in the fairness state select a random slot time totransmit in the fairness period, and nodes in the deferral state select a randomslot time in the deferral period. A node which has won contention while in thefairness state will switch to the deferral state. Nodes in the deferral state whichhave not won contention will enter the fairness state.In a communication system such as that described in the â634 patent, thenumber of deferral and fairness slots can significantly affect system throughput.As mentioned previously, in a system with a large number of wireless nodescontending for medium access at the same time, a larger number of fairness anddeferral slots is desired, whereas fewer slots are desired when only one or twowireless nodes are contending for medium access.For example, if node 102 of Figure l is the only node wanting to senddata at a given time, node 102 would have to choose a slot in the deferralperiod which is at least longer than the fairness period. Thus, if the fairnessperiod was one slot, node 102 could transmit at a minimum of one slot time. Ifthe fairness period was four slots, node 102 would have to wait at least fourslot time periods. However, having one fairness slot would not work in alarger system. For example, if four nodes wanted access to the medium at thesame time, and all ended up in the fairness period, all four would contend inthe same slot and none of the four nodes would win contention. Consequently,none of the four nodes would be able to transmit data.Accordingly, the present invention is directed to a method and apparatusfor monitoring the number of wireless nodes in the communication system atany given time, and adjusting the number of slots accordingly. For example,where a communication system is configured with an access point to which allnodes in the system direct data, the access point can include a master stationconfigured as described above. The access point can then monitor the numberof data transmissions in a given time period. Based on this information, themaster station can configure the number of deferral slots ("M" in Figure 3)-16-?WO 98/128291015202530CA 02265478 1999-03-03PCT/US97/16356and/or the number of fairness slots ("N") in Figure 3) through, for example,the Start of B message.In an access point communication system, where all nodes communicatethrough an "access point" (AP), a node which wishes to transmit data over thecommunication system will send a packet to the access point with its "source"address included. The access point creates a linked list of different "source"nodes from which it receives a packet. This list is sorted by, for example, atimestamp associated with the time at which the packet was received.Therefore, to find out if there was more than one node transmitting in the lasttwo seconds, the access point would examine the timestamp of the secondelement in the linked list to see if it was within the last two seconds. Further,if there were eight elements in the list, the access point can determine if therewere at least eight nodes communicating within the last five seconds byexamining the timestamp of the last element to see if it occurred within the lastfive seconds. This information can then be used to automatically adjust thenumber of fairness and/or deferral slots.As those skilled in the art will appreciate, the number of nodes currentlyactive in a communication system (i.e., over a predetermined period of time),can be evaluated in any known fashion to reevaluate the number of slots. Forexample, the number of slots which provide optimum throughput can bedetermined for each of a different number of nodes in the communicationsystem. This information can then be stored in the access point. Subsequently,the access point can determine the number of nodes currently active in thecommunication system over a current predetermined period of time, and usingthe infonnation stored in the access point, reset the number of slots for the nextpredetermined period of time based on conditions in the current period of time.The period of time during which data transmissions are monitored can be set bythe user.Figure 4 illustrates operation in accordance with an exemplaryembodiment. In step 402, a linked list structure having a predeterminednumber of elements (e.g., eight elements in an exemplary embodiment) isinitialized. This list is used to store the timestamps associated with the-17-?W0 98/128291015202530CA 02265478 1999-03-03PCT/U S97/ 16356predetermined number of source nodes active within the predetermined time.For purposes of the following discussion, a maximum number of slots will beused when at least eight source nodes are active in the communication systemover a five second period.In step 404, the maximum number of slots (e.g., eight) are initialized to4- a normal setting. This normal setting can, for example, be four deferral slotsand four fairness slots.In step 406, a timer used to monitor expiration of the predeterminedperiod of time is set. The predetermined period of time can, for example, be avariable period which is set on the basis of current activity. For example, thepredetermined period of time can be reduced to one second if activity within thecommunication system is relatively light as determined by the number oftimestamps included in the linked list during the originally set five secondperiod. Alternately, if activity in the communication system is subsequentlydetermined to be heavy (as re?ected, for example, by examining the timestampof the tail element in the list to determine if it occurred within the last fiveseconds), then the timer period can be returned to its original value (i.e., fiveseconds). Of course, any number of different predetermined periods of timecan be used, with the predetermined periods of time being selected based oncurrent activity in the communication system.In step 408, an event manager (which can be configured as a processorwithin the access point) monitors source node activity within the predeterminedperiod of time. In step 410, the event manager determines whether a currentlyreceived packet of information corresponds to a source node already included inthe linked list. If so, the event manager moves the source node to the head ofthe linked list in block 412, and updates the timestamp associated with thereceipt of that data packet. Alternately, if the source node is not already in thelinked list, then it is stored in the tail element of the linked list by, for example,copying the source node address and the timestamp associated with receipt ofthat packet in block 414. The event manager then moves the tail element to thehead of the linked list in block 412. For example, if a ninth packet is detected,it replaces the tail element in block 414, and the tail element is then moved to-18-?W0 98I128291015202530CA 02265478 1999-03-03PC'lâ/US97/ 16356the head of the list in block 412, since it constitutes the most recently receiveddata.To set the number of slots used by the communication system, the eventmanager performs a sequence of steps during the expiration of eachpredetermined period of time. More particularly, in decision block 416, theevent manager determines whether a second element in the linked list wasintroduced to the linked list structure within the last five seconds. If not, theevent manager can determine that the communications system is operating at avery light level of operation, and can reduce the number of slots to theminimum via step 418. In an exemplary embodiment, a minimum number ofslots is used when less than two active source nodes exist in the communicationsystem over the predetermined period of time (e.g., one fairness slot and onedeferral slot).An alternate branch of decision block 416 occurs if a second sourcenode element was active within the five second predetermined period of time.In this case, decision block 420 determines whether a tail element of the linkedlist occurred within the last five seconds. If not, then operation proceeds toblock 422, wherein a light mode of operation is determined to exist, such thatan intermediate number of slots is selected for use in the communicationsystem. In an exemplary embodiment, three such slots are used (e.g., twofairness slots and one deferral slot).An alternate branch from decision block 420 occurs if the tail element ofthe linked list structure did occur within the last five seconds. In this case, themaximum number of slots are set to initiate a normal mode of operation (e.g.,eight slots, four of which are fairness slots and four of which are deferralslots).Thus, the number of slots can be configured as follows: (1) if theaccess point detects no more than two active nodes in the predetermined periodof time, two slots (i.e., one fairness slot and one deferral slot) can be used; (2)if the access point sees eight or less active nodes, three slots can be used (i.e.,two fairness slots and one deferral slot); and (3) if the access points sees eightor more active nodes, eight slots can be used (i.e., four fairness slots and four-19-?W0 98/128291015202530CA 02265478 1999-03-03PCT/US97/16356deferral slots). Of course, any number of slots can be used in conjunction withexemplary embodiments of the present invention. Further, any criteria can beused to set the number of slots associated with a given activity level in thecommunication system.As those skilled in the art will appreciate, the categories of activity usedto define the number of slots will necessarily be a function of the predeterminedtime over which communication system activity is monitored. Accordingly, anempirical approach can be used to determine an appropriate period of time overwhich to monitor activity so that the different categories of slots used will havea meaningful effect on system throughput.While the foregoing discussion has focussed on the monitoring of actualdata transmissions in a predetermined period of time to measure communicationsystem activity and dynamically set the number of slots used, those skilled inthe art will appreciate that numerous variations exist. For example, rather thanmonitoring data transmissions through a central point of a communicationsystem, such as an access point that bridges two different types ofcommunication mediums (e.g., a wireless medium and a wired medium), anyinformation which serves as an indication of communication system activity canbe used. For example, rather than monitoring actual data transmission, therequestâto-send messages can be monitored and used to adjust the number ofslots. The monitoring of RTS messages avoids the need to include an accesspoint through which all data transmissions occur in the communication system.Further, those skilled in the art will appreciate that the number of slotsused can be set as desired by the user. Further, while both fairness anddeferral slots as described in the â634 patent have been incorporated in thediscussion of an exemplary embodiment of the present invention, those skilledin the art will appreciate that a distinction between fairness and deferral slots isunnecessary. That is, the number of slots used, regardless of whether they arefairness or deferral slots can be dynamically adjusted. Further, where fairnessand deferral slots such as those described in the â634 patent are used, a ratio of1:1 between these different category slots need not be maintained. â Rather, theratio of fairness slots to deferral slots can be dynamically adjusted either as the-20..?CA 02265478 1999-03-03W0 98/12829 PCT/US97/16356sole dynamic adjustment, or in combination with an adjustment to the totalnumber of slots.It will be appreciated by those skilled in the art that the presentinvention can be embodied in other specific forms without departing from the5 spirit or essential characteristics thereof. The presently disclosed embodimentsare therefore considered in all respects to be illustrative and not restricted. Thescope of the invention is indicated by the appended claims rather than theforegoing description and all changes that come within the meaning and rangeand equivalence thereof are intended to be embraced therein.-21-

Claims

WHAT IS CLAIMED IS:

1. Apparatus for controlling wireless communication among plural nodes in a frequency hopping communication system comprising:
means for monitoring a number of transmissions which occur within the communication system; and means for dynamically changing a number of slots for which nodes of the communication system contend for access to a wireless communication medium of the communication system, said dynamic changing being performed as a function of said number of monitored transmissions.

2. Apparatus according to claim 1, wherein said monitoring means detects a number of data packet transmissions sent within a predetermined period of time.

3. Apparatus according to claim 1, wherein said monitoring means detects a number of request-to-send messages sent within a predetermined period of time.

4. Apparatus according to claim 1, wherein said monitoring means further includes:
means for maintaining a linked list of nodes which have transmitted information within a predetermined period of time.

5. Apparatus according to claim 4, wherein said linked list includes timestamp information regarding a time at which each of said nodes transmitted information, said linked list being sorted according to said timestamps.

6. Apparatus according to claim 1, wherein said means for dynamically changing is performed as a function of said number of monitored transmissions received in a predetermined period of time, said predetermined period of time being variable based on activity of the communication system.

7. Method for controlling wireless communication among plural nodes in a communication system, comprising the steps of:
monitoring a number of transmission which occur within the communication system; and dynamically changing a number of slots for which nodes of the communication system contend for access to a wireless communication medium of the communication system, said step of dynamically changing being performed as a function of said number of monitored transmissions.