US 20050147075 A1
An inventive method provides optimum topology for a multi-antenna system dedicated to higher throughput/capacity by bundling the Point Coordination Function (PCF) operation in infrastructure mode of the current and/or enhanced IEEE MAC with PHY specifications that employ some form of coherent weighting based on CSI at the transmitter in conjunction with the corresponding optimum receiver detection based on CSI. Specifically, CSI is measured from a control message, so data messages and control messages are separated. In the contention period of IEEE 802.11, the RTS/CTS exchange is used for CSI and the data message is sent following the CTS message. In the contention free period, a poll by the PC is separated from a data frame, which gives the polled station the first opportunity to send a data message. This change in topology results in various changes to the frame exchange format in the CFP for various scenarios of data and control messages to be exchanged.
25. In a wireless local area network wherein a first network entity transmits to a second network entity a packet having a guard interval preceding one of a data signal and a training sequence, the improvement comprising:
the first network entity measuring channel state information CSI for the channel between the first and second network entity;
the first network entity selecting a length of the guard interval based on the CSI; and
the first network entity sending the packet with the guard interval of length selected based on CSI.
26. In the wireless local area network of
the first network appending to a tail end of the packet an iterative decoding signal extension.
27. In the wireless local area network of
28. A method for transmitting a packet in a wireless local area network WLAN, comprising:
determining, at a first network entity, channel state information CSI for a channel between the first and a second network entity in a WLAN;
selecting, at the first network entity, a guard interval length from among at least two lengths based on the determined CSI; and
sending a packet with the guard interval of the selected length from the first network entity over the WLAN.
29. The method of
encoding the packet with a capacity enhancing code, and including in the packet an iterative decoding signal extension disposed at a tail end of said packet.
30. The method of
31. The method of
32. The method of
33. The method of
34. The method of
35. A network entity for communicating over a wireless local area network WLAN comprising:
a receiver for receiving a message from an entity of a wireless local area network WLAN;
a memory for storing at least two different guard interval lengths, each associated with a different channel state information CSI;
a processor, coupled to the receiver and the memory, for determining CSI from the received message, and for selecting from the memory one guard interval length associated with the determined CSI;
a transmitter having an input coupled to an output of the processor, and an output for coupling to an antenna, said transmitter output for outputting a packet comprising a guard interval of the selected length.
36. The network entity of
37. The network entity of
38. The network entity of
39. The network entity of
40. The network entity of
41. The network entity of
42. The network entity of
43. A device for communicating over a wireless local area network WLAN comprising:
a receiver for receiving a message from a node of a WLAN;
means for determining channel state information CSI from the received message;
means for determining a guard interval length from at least two lengths based on the determined CSI; and
means for transmitting a packet comprising a guard interval of the selected length.
44. The device of
45. The device of
46. The device of
The present invention claims priority from co-pending Provisional U.S. Patent Application No. 60/460,553, filed with the U.S. Patent Office on Apr. 4, 2003.
The present invention relates broadly to Wireless Local Area Networks (WLANs) and specifically to a topology for multi-channel wireless time division duplex (TDD) systems so that channel state information (CSI) may be acquired and used to optimize data throughput.
Highly functional computers have been interconnected with one another in what is termed a local area network (LAN) to enable users of individual computers within a predefined set to share files with one another. Traditional hardwired LANs are being superceded by wireless LANs (WLANs) as WLANs realize increased capacity. Data protocols for WLANs are generally organized into layers or levels of the communication system, each layer facilitating interoperability between various entities within the network.
The Institute of Electrical and Electronic Engineers (IEEE) standard for WLANs, IEEE 802.11, provides protocols for a physical (PHY) layer and a Medium Access Control (MAC) layer, shown in block diagram form at
The MAC layer 25 is a set of protocols that maintain order in the use of the shared bandwidth or medium, and the 802.11 standard specifies two modes of communication: a compulsory Distributed Coordination Function (DCF) 26, and an optional Point Coordination Function (PCF) 27. A Basic Service Set (BSS) 31 is shown in
Avoiding collisions (simultaneous transmissions) between stations in a BSS is complicated by the fact that while a wireless station is transmitting, it cannot monitor the transmission medium (the channel or channels) for other traffic that may interfere with its own transmissions. For example, one problem arising from the inability to listen while transmitting in WLANs is termed a “hidden node”. Assume stations A, B and C in a BSS are disposed as in FIG. 1B, with B physically located between A and C. If stations A and C cannot communicate directly with one another due to distance, multipath fading, or some other reason, stations A and C are hidden from one another. Absent some collision control scheme, station A may listen to the channel, sense it is clear, and transmit a packet to station B. Whether or not station C is transmitting to B is unknown to A, except through coordination by the PC. Simultaneous transmissions from stations A and C to station B would result in collision and lost transmissions, since all stations in a BSS 31 communicate over the same channel.
DCF seeks to minimize collisions by prioritizing stations waiting to transmit based on a time delay basis. In DCF, each station 32 with a data message to transmit contends for the next available slot on the BSS channel during what is termed a contention period CP 29. Time delays for various stations have a random component, but procedures ensure a waiting station moves up in priority the longer it waits. Details of the DCF prioritization protocol are described in detail below. Once a station sends its data message, which is included in a MAC Service Data Unit (MDSU), it must contend with all other waiting stations for another available slot. PCF is provided to avoid the situation where time-sensitive data from one station cannot be assembled into one MDSU, which is constrained to a maximum length. For example, station A may wish to send an audio or video clip that spans three MDSU's to station B, but contending for a separate transmission slot for each of the MDSUs would potentially result in the clip being undecipherable. While a relatively large buffer in the receiving station may store and re-assemble the separately received clip portions after a not insignificant delay, that option is generally not seen as viable in the long term due to the dual constraints of low power consumption and small physical size of wireless stations. When implemented, PCF takes priority over DCF in that a contention free period (CFP) 28 is established whereby station A may send its data messages without contending for a time slot. During the CFP 28, other stations stand by and await either a poll by the PC during the CFP 28 or a contention period (CP) 29 in which the various stations contend for a slot as in DCF above. Additional details of PCF are provided below.
Historically, the development of WLAN systems, and wireless systems in general, have taken two paths, one focused on specifications for the PHY layer and the other for the MAC layer. For example, the IEEE 802.11(e) task group is developing MAC layer enhancement to improve MAC layer throughput regardless of physical layer throughput. The IEEE 802.11(g) task group has developed a physical layer specification that facilitate data rates of 20+ megabits per second (Mbps) in the 2.4 GHz. Range, but must keep MAC layer changes to a minimal. Though both working groups operate concurrently, in practice there appears little interaction between the two groups. Advantages that may be gained by a more holistic approach are never recognized by the groups' single-layer focus.
Recently, the IEEE has approved a High Throughput Study Group (HTSG) for 802.11, whose charter is to provide higher throughput than enabled by current IEEE 802.11 standards. The High Throughout Task Group (HTTG) will develop the actual standards, which appears to be the first time that modifications to the MAC and physical layers will be developed coherently since the division of those layers. A recent study showed that the current IEEE MAC and physical layers is limited to a throughput of 0.2 Mbps per 1000 byte packet per operational mode. Existing 54 Mbps modes therefore have approximately 28 Mbps throughput for a 1000 byte packet. Maintaining the same ratios, then a 108 Mbps data rate yields a throughput of 56 Mbps for a 1000 byte packet.
It is well-known that optimum capacity is achieved when Channel State Information (CSI) is known and used at both the transmitter and receiver, and that MIMO systems (multiple input/receive antennas and/or multiple output/transmit antennas) provide a substantial increase in capacity as compared to more traditional systems employing a single antenna on all transceivers. For example, knowing CSI enables a transmitter to parse data among different channels in a manner that takes advantage of the entire channel capacity on each channel, rather than allowing the time-sensitive bandwidth to be not fully used. Some communication standards such as Code Division Multiple Access (CDMA) reserve a feedback channel to provide CSI to the transmitter. Unfortunately, CSI via a feedback channel is imperfect due to feedback delays and changing channel characteristics. Regardless, the 802.11 standard does not entail a feedback channel, there are no physical layer specifications in 802.11 that are based on CSI, and some researchers believe the lack of CSI in the standard prohibits the adoption of a feedback channel in future versions of 802.11.
Thus, there is a need in the art to provide an optimum throughput/capacity topology for multi-antenna wireless systems that imposes changes that are backwards compatible with current WLAN stations.
Fortunately, there are resolutions to this problem that are embodied in the present invention. As mention above, there are no physical layer specifications in the IEEE 802.11 standard that are based on CSI at the transmitter. Operation of the Contention Free Period (CFP) is described in the IEEE 802.11(e) draft standard, herein incorporated by reference. Depending on the physical layer standard 802.11(a), 802.11(b) or 802.11(g), the CFP modulation is derived from one of their operational modes.
A system according to an embodiment of this invention provide the optimum topology for a multi-antenna system dedicated to higher throughput/capacity by bundling the Point Coordination Function (PCF) operation in infrastructure mode of the current and/or enhanced IEEE MAC with PHY specifications that employ some form of coherent weighting based on CSI at the transmitter in conjunction with the corresponding optimum receiver detection based on CSI.
In one embodiment of the present invention is a method of communicating over multiple sub-channels of a WLAN. The method includes sending a control message that is not combined with a data message from a first network entity to a second network entity. The control message may be, for example, a CTS message during the CP or a poll during the CFP, but in any case the control message is to facilitate sequencing of wireless transmissions among at least two entities in a wireless network. In the inventive method, the control message is received at the second network entity, which uses it to obtain channel state information CSI. The CSI is used to determine the capacities of at least a first and second sub-channel of the wireless network, and to determine which has the greater capacity. A data message to be sent is divided into at least a first and second data message segment, wherein the relative sizes of the segments are based on the relative capacities of the sub-channels. The division itself is preferably via an eigenmode or water-filling known in the art to exploit varying capacities of sub-channels. When the first sub-channel is determined to have the greater capacity, the first data message segment will then define a greater size than the second data message segment. Further in the method and in response to receiving the control message, the second network entity sends the first data message segment over the first sub-channel, and the second data message segment over the second sub-channel of the wireless network. In this manner, CSI is obtained and used to send the segmented data message, though not necessarily the control messages.
In a particular embodiment, the first network entity is a point coordinator PC of a wireless network basic service set BSS operating during a contention free period CFP, the control message is a poll of the second network entity, and the PC may respond with an ACK message combined with a data message for the first network entity. Preferably, where the PC sends a poll of a third network entity during the same CFP as the poll of the second network entity, and the PC fails to receive a response from the third network entity within a first time period such as a SIFS, the PC then polls a fourth network entity within a second time period such as a PIFS that is no greater than twice the first time period. Where the PC receives from a network entity an ACK message combined with a data message, the PC may respond with an ACK message combined with a separate control message that signals an end of a contention free period. In the 802.11 standard, for example, such a message from the PC would be a combined ACK and CFP-End message.
Further according to another aspect of the present invention, when the method is executed during a contention free period CFP, and the first network entity is a point coordinator PC and the control message is a first poll of the second network entity, there exists an instance where a polled station does not respond to its poll. To avoid confusion with the terms above, assume an initial poll of an initial network entity or station occurs prior to the poll of the second network entity or station. Prior to sending a control message without a data portion from the PC to the second network entity, the method preferably also includes sending from the PC an initial poll without a data message to an initial network entity. Upon the PC failing to receive a response to the initial poll from the initial network entity within a first time period such as a SIFS, the PC then preferably sends, within a second time period such as a PIFS that is greater than the first time period, either a data message to the initial network entity or the first poll of the second network entity as described above.
The present invention may also be adapted for station-to-station data communications during the CFP. Where the method as summarized above is executed during a CFP, the data message in its various segments is sent over the sub-channels from the second network entity to a third network entity that is not a point controller PC. In that instance, the method further includes the third network entity sending to the second network entity an ACK message within a first time period, in response to receiving the data message segments. The PC may then send, within a period of time following the ACK message from the third entity that is less than twice the first time period, either a poll to a network entity, or a data message to the second network entity that is divided into data message segments based on CSI that is measured from at least one data message segment sent from the second network entity to the third network entity. If the PC is to allow the second and third stations to exchange multiple data messages between them, the PC will wait a PIFS before transmitting. If the PC is to allow only one cohesive data message from the second to the third entity, it need wait only one SIFS after the ACK message from the third to the second entity, or one PIFS following the data message from the second to the third entity.
In the above method, at least one of the network entities is preferably a mobile station such as a mobile phone. The term mobile station as used herein includes any portable electronic device that has a telephonic capability, such as cellular phones, portable communicators, PDAs with telephonic capability, and further includes the various accessories to the above that expand the capabilities or functionality of the mobile station with which they are coupled.
According to another embodiment of the present invention is a method of communicating data over a wireless network according to an IEEE 802.11 standard as it exists as of the priority date of this application. In this embodiment, the improvement to the 802.11 standard includes separating by at least one Short InterFrame Space SIFS a poll and a data message sent by a point controller PC while in a contention free period CFP. This allows data messages sent from the PC to be transmitted with the benefit of knowing CSI, with at least one possible exception noted below.
Preferably, CSI is also obtained during the contention period CP during a Request-to-Send/Clear-to-Send RTS/CTS exchange. In that instance, CSI is used to determine relative capacities of at least a first and second sub-channel to parse a data message from a station sending the RTS to a station sending the CTS. Specifically, a data message from the RTS-sending station is parsed into at least a first data message segment defining a first size and a second data message segment defining a smaller second size. The relative segment sizes are based on relative capacities of a first and second sub-channel as determined by the measured CSI. The larger first data message segment is sent over the higher capacity first sub-channel and the smaller second data message segment is sent over the lower capacity second sub-channel. Parsing of the overall data message is based on relative sub-channel capacity as determined by the measured CSI, such as by eigenmode or water-filling techniques known in the art.
Considering again the CFP, this embodiment of the present invention preferably restricts the PC to sending only one of five possible messages: a poll; a data message parsed according to measured CSI and transmitted among at least two sub-channels; a data message so parsed and transmitted combined with an ACK message; a CFP-End message; and a CFP-End message combined with an ACK message. Conversely, 802.11 currently allows a data message to be combined with a poll message, and does not provide that an ACK can be combined with a CFP-End message since there appears no opportunity for the latter to ever need be combined as the standard currently exists. Preferably, the PC can combine a data message only with an ACK message, else the data message may not be combined with any other message.
Preferably, the PC is allowed to send a data message without valid measured CSI to a station only upon non-receipt of a response from that same network entity to its poll within one SIFS. Most preferably, the PC can only send a data message with either valid measured CSI or estimated CSI.
Where the PC and the polled station each have a data message to send, one difference of the present invention as compared to the 802.11 standard is that the polled station is preferably allowed to send its data message first. Preferably, between the time the PC polls the station and the time the PC may next transmit, the polled station may send a data message to another station (that is not the PC) without using measured valid CSI for the channel between the polled station and the another station. In this instance, the another station is allowed an opportunity (one SIFS) to send an ACK message to the polled station prior to the time the PC is next allowed to transmit.
Another aspect of the present invention is a network entity for communicating over a wireless local area network, such as a mobile station, a point controller, an access point, or any other entity on the WLAN. The network entity includes a receiver for receiving over at least two sub-channels a control message from an entity of a wireless local area network. The control message is preferably a CTS message or a poll. The mobile station further has a processor for determining a capacity of a first sub-channel and a capacity of a second sub-channel based on channel state information CSI measured from the control message. It further includes means for parsing a data message into at least first and second segments based on the relative determined capacities of the first and second sub-channels. To best exploit the multi-channel capability in both transmit and receive functions, the mobile station has a first and second antenna having inputs coupled to an output of the means for parsing. The first antenna is for transmitting at least the first segment over the first sub-channel and the second antenna for transmitting at least the second segment over the second sub-channel. In certain embodiments, there may be a crossfeed between antennas with differential weighting for each data message segment so that each segment is actually transmitted over each sub-channel, and increased capacity is realized by the differential weights assigned to each segment.
These and other features, aspects, and advantages of embodiments of the present invention will become apparent with reference to the following description in conjunction with the accompanying drawings. It is to be understood, however, that the drawings are designed solely for the purposes of illustration and not as a definition of the limits of the invention.
The present invention is better understood in light of the following drawings.
In the 802.11 standard, a Point Controller (PC) coordinates prioritization during the contention free period CFP 28. The PC is functionally within the Access Point (AP) 33 of a BSS 31 and is usually physically collocated with it, so the term AP 33 is used herein to indicate either or both the AP 33 and PC. A station 32 may serve as the AP 33 and the CP.
A superframe 35 begins with a beacon frame 36 transmitted by the PC, regardless of whether PCF is active or not. The beacon frame 36 is a management frame that provides timing and protocol related parameters to the stations. Each beacon frame 36 also announces when the next beacon frame will be transmitted, so that all stations 32 are aware of superframe lengths. To enable PCF 27 to take priority over DCF 26, the PC transmits the beacon frame 36 after a PCF Interframe Space (PIFS) 37 (about 25 μs) following the end of the last superframe 35. Because the PIFS 37 is shorter than a DCF Interframe Space (DIFS, about 34 μs) that the DCF 26 must wait following the end of a superframe 35, PCF 27 can take priority. A Short Interframe Spacing (SIFS) 38 spans about 16 μs and is the amount of time a station 32 is allowed to reply to the PC. Each station 32 within the BSS 31 resets a Network Allocation Vector (NAV) 41 based on the beacon frame 36. In
After the beacon frame 36, the PC delays one SIFS 38 and may send any of the following: a data-only frame, a data+poll frame 42, a poll-only frame, or a CFP-end frame. The PC maintains a list of stations for which it has data, and typically polls those stations first in order to piggyback that data with its poll of the station. Referring to
After receiving the data+ACK frame 43 from the first station (U1+ACK), the PC waits one SIFS and polls another station (or the same station). In the event the previous first station sent its data (U1) to the PC, the PC will piggyback an ACK for that first station in the data+poll it sends to a second station in a data+poll+ACK frame 44 (D2+ACK+Poll, data and poll directed to the second station, ACK directed to first station). In
One drawback with the prior art, at least in certain circumstances, is that the polling frames and the data frames from the PC may be combined into a single frame (data+poll 42 or data+ACK+poll 44). At the time of that combined frame transmission, the PC does not know the channel state between it and the intended station. While channel state may not change significantly over a single CFP repetition interval 35 when used in a wired network, channel states change much more rapidly in WLANs. To increase capacity over a fixed bandwidth, multiple sub-channels are preferably used such as in a single input/multiple output (SIMO) communication system, a multiple input/single output (MISO) system, or most preferably a multiple input/multiple output (MIMO) system. Any of these are referred to hereafter as a MIMO system unless otherwise stipulated. The multiple sub-channels of a wireless MIMO system are each subject to rapid changes due to fading, multipath, etc., so wireless MIMO systems need to know the state of the different sub-channels to send different data portions over the strongest channels, or to partition the data to be sent into sizes that maximize the respective capacities of the various sub-channels as those sub-channels exist at the time of transmission. When the PC polls a station, it has not yet received any feedback from that station by which to measure the true channel. Since the sub-channels change rapidly, it is highly unlikely that the coherence interval (the interval over which the measured state of the channel does not change significantly) spans an entire CFP repetition interval 35. Said another way, any measurements of the channel made in one CFP 28 are unlikely to be valid estimates of the channel during the next CFP 28. Sending a data message combined with a poll necessarily implies sending the data either regardless of channel quality or with invalid (i.e., outside the coherence interval) estimates of the channel. Either option is a waste of bandwidth as compared to maximum capacity theory. Among other aspects, the present invention modifies the specific frame exchange of
On first glance, it appears the exchange of frames of
As an alternative to the scenario described for
In any of the above instances, any of the PC or stations may have more than one frame with data to send. Due to the potential size of the data frames and the speed with which the channel may vary over time (the length of the coherence interval), it may be necessary in one instance that the sender re-acquire CSI from the last transmission of the intended recipient, and in another instance it may have negligible effect on data throughput that the sender re-use the originally measured CSI. So long as the frames in question are sent within the coherence interval established when CSI is measured, then CSI is considered valid whether or not is was measured based on a frame received immediately preceding the next frame to be sent.
The above description pertains to the CFP 28 wherein the PC controls which station in an infrastructure network may next transmit. Following is a description as to how the present invention may be used within the contention period 29 following the CFP 28. Since the CFP 28 exists only while in the point coordination function 27, operation within the CP 29 is within the base DCF 26 layer of MAC 25 and is detailed at
DCF lies directly on the PHY layer 21 and is based on Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) protocol, because wireless stations cannot listen for collisions while transmitting. As known in DCF, when a station has a frame with data to be transmitted, it first listens to ensure no other station is transmitting over the prescribed channel and transmits only if the channel is clear for a set period of idle time, termed a DCF-interframe space (DIFS) 38 that is longer than a PIFS. If the channel is busy, the station instead chooses a random “backoff factor” which determines a delay period 58 wait until it is allowed to transmit its data. During periods in which the channel is clear, the transmitting station decrements its backoff counter to shorten the delay period 58 so a delayed station gradually gains a higher priority to transmit. When the backoff counter reaches zero and the channel is clear for the duration of a DIFS 38, the station may transmit its frame with data. Since the probability that two stations will choose the same backoff factor is small, collisions between data frames from different stations are minimized.
When a particular station's backoff counter reaches zero and it senses the channel is clear for an entire DIFS 38, that station, termed the source 52 or transmitting station, first sends out a short ready-to-send (RTS) frame 53 containing information on the length of the frame with data to be transmitted. If the intended destination 54 to which the RTS 53 is directed hears it, the receiving station 54 responds with a short clear-to-send (CTS) frame 55. Only after this exchange does the source 52 send its data frame 47 during the CP 29. When the destination 54 receives the transmitted data frame 47 successfully (as determined in 802.11 by a cyclic redundancy check CRC), the receiving station (or PC) transmits an acknowledgment (ACK) frame 48. This back-and-forth exchange is necessary to avoid the “hidden node” problem previously explained. If the receiving station 54 has a data frame 47 to send, it must contend for a transmit slot as above and cannot piggyback data onto its ACK frame 48. During this process, other stations 56 defer transmission access 57 until they sense the channel is clear for a DIFS plus their backoff factor.
The present invention exploits the RTS/CTS interchange to provide valid CSI to at least the source 54 for use in transmitting the data frame 47. The benefits of the destination 54 using CSI obtained from the RTS/CTS exchange for use in transmitting the ACK only frame 48 are relatively minor as that frame is small. Since each station is at differing times both a source 52 and a destination 54, the means to implement the present invention are already in place and can be used for the ACK only frame 48, even if its practical effect is merely to send an unparsed ACK frame 48 over the most robust of the available sub-channels.
There is another opportunity within the 802.11 standard by which a station may obtain valid CSI for the channel over which it intends to transmit. A listening station, such as the other station 56 of
The minimum criteria for optimum transmission topology for wireless time division duplex TDD networks are:
To achieve the capacities possible with the present invention, the transmitter should employ some weighting mechanism to assign frames, packets, fragments, or whatever division of the entire package to be transmitted to various sub-channels based on the measured quality of those sub-channels. Eigen-mode or waterfilling is one technique known in the art to do so, described mathematically below. For ad hoc networks and infrastructure networks during the contention period, the RTS/CTS exchange may be used. During the contention free period, the revised frame exchange described above may be employed to achieve valid CSI. In either case, the coherent weighting is done at the PHY layer 21, so the present invention modifies both the MAC and PHY layers.
Frame Efficiency as used in Table 1 is the time required to transmit the information portion of packet divided by the total on air time for packet. Thus, the overall capacity is found by multiplying the frame efficiency by the capacity/throughput, which are shown in Table 2 below:
The capacity requirements are computed as raw data rate/12 Msymbols/sec/Frame efficiency to yield the target throughput/capacity at the MAC SAP layer. The theoretical best performance for these capacity requirements can be read from
Eigen-mode transmission as noted above is described as follows. Let the singular value decomposition of H be H=UΣV where U and V are unitary matrices and Σ be a diagonal matrix With positive real values on the diagonal elements representing the singular values of the channel. If the transmitted vector r is pre-multiplied by V in the transmitter and received vector is post multiplied by UH in the receiver, i.e., Vr UH=V (Hx+n) UH=Σx+m, where m=Vn*UH and there is no noise amplification and remains spatially white.
Because a single MAC layer must interface with disparate PHY layers, the 802.11 standard uses an additional protocol layer termed the Physical Layer Convergence Protocol (PLCP) disposed between them that is defined differently for each transmission method. The PLCP adds a preamble and a header (each of various sizes) to a PLCP Service Data Unit (PSDU), which is the portion of the complete transmission frame (PPDU or PLCP Protocol Data Unit at the PHY layer) that carries the actual data to be transmitted between stations or between the point controller PC and a station.
While there has been illustrated and described what is at present considered to be a preferred embodiment of the claimed invention, it will be appreciated that numerous changes and modifications are likely to occur to those skilled in the art. It is intended in the appended claims to cover all those changes and modifications that fall within the spirit and scope of the claimed invention.