Patent Application
20120051273
The present invention relates to a novel multiple access method, and to apparatus for implementing the method.
The operational cost of low data rate mobile geostationary (GEO) satellite systems is relatively expensive when compared to fixed or transportable counterparts due to their low gain antenna, limited transmit power and extensive fading margin. For such systems more satellite resources, such as bandwidth and power, should be made available to accommodate poor propagation conditions which usually become worse as the user terminals operate at the edge of coverage, viewing the satellite at lower elevation angle (e.g. as low as 5°). One approach is to limit the service area so that only high elevation satellite links are allowed (e.g. higher than 20°) ensuring good propagation conditions, but this clearly does not satisfy the requirements for provision of services at high latitudes such as Northern European countries. One solution is to use higher gain antennas at the terminals and to include some mechanism to steer the antenna beam towards the satellite. Another solution is to design waveforms with time-diversity to provide protection against short-duration deep multipath fades but this has the disadvantage that low-latency services such as voice cannot be supported. A further solution is the use of space-diversity antenna subsystems, but while this may be suitable for some types of mobile installations, the possible applications are somewhat limited.
An example of a modern mobile satellite system is the Inmarsat (RTM) Broadband Global Access Network (BGAN) which has been deployed to provide high-speed IP data services to very compact transportable terminals using a global network of multi-beam GEO satellites at L-band. Recently, the BGAN system has been enhanced to extend the mobile terminal portfolio to include land mobile, maritime and aeronautical services such as Swiftbroadband (SBB). Inmarsat is continuously enhancing the BGAN system to support new services such as push-to-talk netted voice-over-IP and other interactive group data applications. Such applications involve bursty traffic from user to user interaction with very tight latency requirements. The low-latency of bursty traffic requirement is difficult to address efficiently using a conventional TDMA waveform such as employed in the existing BGAN air interface. New radio resource management algorithms, including modifications to the signalling and the behaviour of both Mobile Terminals and Satellite Access Stations are likely to be introduced to support such service capabilities in the near future. The introduction of extensions to the existing BGAN physical layer able to support unscheduled asynchronous transmissions from Mobile Terminals (MT) in the return link, would both minimize system complexity while enhancing system performance and capacity for such applications.
An example of mobile satellite service employing compact low gain antennas and requiring high reliability over the widest possible area is air traffic management (ATM). The forecast increase in air traffic has stimulated research into advanced satellite based systems for air traffic management. Satellite-based systems are already used for provision of aeronautical safety services in Oceanic airspace, and offer significant cost advantages as a secondary communications path to improve overall system availability for Continental airspace, but systems employing low gain antennas on the aircraft to provide higher data rates are still to be defined and the expected performance remains to be validated in an operational environment. Current systems for ATM in Oceanic airspace such as Inmarsat Aeronautical Mobile Satellite Service (AMSS) and the Japanese MTSAT employ mostly intermediate and high gain antennas which are large and expensive when small civil aircraft are considered.
State-of-the art satellite systems such as Inmarsat BGAN and DVBS/RCS use allocation schemes which operate in both frequency and time domains (e.g. MF-TDMA) as shown in
According to one aspect of the invention, there is provided a method according to claim 1. According to another aspect of the invention, there is provided a system according to claim 39. According to another aspect of the invention, there is provided a transmitter according to claim 76.
According to embodiments of the invention, joint access to spectrum together with multi-burst decoding at the receiver can significantly increase the system capacity by increasing the energy density in the channel, while simultaneously improving service latency and reducing the overheads associated with network control functions. Complexity may be transferred from the signalling and control infrastructure into the receiver, this being equipped with enhanced signal processing techniques making possible to extract information streams from composite signals. The use of high redundancy turbo-coding technology in the return direction and the associated novel receiver architectures may be used to enhance the performance of the reserved and random-access signalling.
In contrast to the current BGAN return link, unscheduled transmissions from vehicular, maritime and aeronautical (including helicopter) terminals may reuse the satellite resources. All these terminals may use the same time and frequency slot for their transmission. The new type of Satellite Access Station (SAS) receiver processes the received composite signal from all these transmissions which are corrupted by high mutual interference and detects each transmission. Therefore, the satellite spectrum is not shared but instead is reused by a number of terminals and this feature increases dramatically the system capacity, reduces the system latency and simplifies the signalling protocols.
In contrast to a conventional DS-CDMA system, the proposed BGAN system enhancement uses highly redundant turbo coded narrowband transmissions instead of CDMA techniques and their associated spreading sequences. For this reason the proposed multiple access scheme may be referred to as a Turbo Code Division Multiple Access scheme (TCDMA) and uses narrowband multiuser detection. The TCDMA signals in space are totally compatible with the legacy BGAN system and at the same time are extremely robust against multipath fading from different mobile satellite propagation channels and interference, originating from similar transmissions which share simultaneously the same frequency channel. In contrast to the traditional CDMA, the capacity of a TCDMA system increases as the simultaneous transmissions differ in power. Therefore, there may be no need for power control signalling. Small time and frequency errors between the simultaneous transmissions and multipath fading are different for each mobile terminal helping the receiver to discriminate the various transmissions. The system capacity increases as the coding rate decreases and the transmissions become uncoordinated. This is a significant feature of TCDMA in contrast to traditional approaches.
Another attractive feature is that robustness of each transmission increases as the traffic loading decreases and this may be provided without any involvement from signalling protocols. Therefore, the proposed TCDMA multiple access scheme is very suitable for sporadic unscheduled traffic from mobile terminals equipped with low gain antennas. Some of the benefits of TCDMA configuration are:
enhanced random access capacity: minimizes signalling delay, maximizes capacity for random access and simplifies protocols for sporadic transmissions;
enhanced user traffic capacity: scheduled and unscheduled user traffic can share return slots;
reduction of signalling latency: signalling protocols that introduce considerable latency due to the satellite round trip may be avoided; and
simplification of network protocols: since slots can be shared, less radio resource reservation and signalling are necessary.
The complexity of a multi-burst receiver is significantly higher than that of a single channel receiver. Complexity is transferred into the receiver incorporating new functions which can extract information from a composite signal. The complexity of a multi-burst receiver is higher than conventional single-burst receivers. In general, as the receiver becomes more complex, its performance improves. However, there is a practical limit on how complex the receiver can be.
Embodiments of the invention may provide a minimum complexity architecture that retains most of the performance benefits mentioned above. The architecture may bridge the gap between single burst and multi-burst receivers. Furthermore, this architecture appears to perform very well with a realistic operational scenario and with the potential of approaching the theoretical capacity of the multiple access channel.
State-of-the-art fixed (e.g. DVBS2/RCS) or mobile satellite systems such as the Broadband Global Access Network (BGAN) and its aeronautical version Swiftbroadband system (SBB) use mostly orthogonal allocation schemes in frequency and time to simplify the transmission and reception technologies. The SBB return link is a typical example, where each aeronautical, terminal transmits a burst of data always within a unique time/frequency slot. This, however, puts organizational and computational burden on the network controller. Embodiments of the invention provide a new type of multiple access which may improve the bandwidth efficiency and reduces the latency of modern satellite systems using highly redundant turbo coding and new receiver architecture.
One embodiment comprises a multi-burst receiver with a multiuser detector at the satellite access station capable of detecting more than one burst per time and frequency slot.
The scheme according to embodiments of the invention may use highly redundant turbo coded narrowband transmissions. For this reason the proposed multiple access scheme is named Turbo Code Division Multiple Access scheme (TCDMA). Joint accessing of spectrum using multiple-burst transmission and multiuser decoding strategies has the potential of significantly increasing the system capacity while reducing the overhead associated with network control functions and various signalling latencies. Complexity is transferred into the receiver, equipped with signal processing functions which can extract information streams from composite signals. With the rapid advances of VLSI technology multi-burst reception appears within reach of modern VLSI chips. The multi-burst receiver includes several sub-receivers and sub-decoders all operating several times until successful detection of the incoming bursts is achieved. The results reveal that the proposed technology nearly doubles the throughput over the conventional single burst receiver. In the random access mode, the application of multi-burst receivers may also eliminate collisions which in turn, may reduce the signalling latency of mobile terminals requesting system resources. There may be a significant increase in receiver complexity but the benefit is also very significant: the system access capacity increases with the number of successfully detected bursts.
Specific embodiments of the present invention will now be described with reference to the accompanying drawings, identified below.
Joint Multiple Access Scheme
In a TDMA system in an embodiment of the invention, as shown in
This scheme can significantly increase system capacity by increasing the energy density in the channel, while simultaneously reducing service latency and reducing the overheads associated with network control functions. Complexity is transferred from the signalling and control infrastructure into the receiver, this being equipped with enhanced signal processing techniques making possible to extract information streams from composite signals.
System Model
A system model in the embodiment is shown in
The bursts have coding rates of R=1/2 or less, typically from R=1/3 down to R=1/9, and are generated by a parallel concatenated turbo encoder. The encoder may be defined by an extension of the CCSDS standard, for example as described in ‘TM Synchronization and Channel Coding’, CCSDS 131.0 B-1, September 2003. The turbo interleaver is as specified for the BGAN air interface and it is an improved version of the CCSDS standard optimized for small packet sizes, as described for example in P. Fines, E. Christofylaki, S. Papaharalabos and P. Febvre, “Low Rate Turbo Code Extensions and Modem Design for High Reliability Satellite Links”, ESA TTC 2007.
The turbo code performance depends on the coding rate and the data size and for R=1/5 and 504 data bits its AWGN performance is shown in
The system may control the coding rate of the transmitters according to the anticipated or measured loading of the traffic channels, and/or the burst error rate measured at the receiver. The coding rate may be same for all transmitters, or may be selected on a per-transmitter basis, for example according to the measured burst error rate for the respective transmitter. The coding rate may be varied by transmitting periodic coding rate control signals addressed to individual transmitters, or broadcast to all transmitters.
Each transmission travels via a different propagation channel PC1 . . . PCn which introduces random propagation characteristics such as different delay, attenuation, frequency shift, multipath fading and/or phase noise. The system has no specific means for ensuring signal orthogonality. However, it should be noted that each channel introduces some signature that is likely to be independent or different for each user by affecting at least some (if not all) of the previously mentioned signal parameters. This unique signature allows the bursts to be decoded separately at the receiver. The propagation channels also add noise to the transmissions, modelled in
In the embodiment, the receiver comprises a multi-user receiver comprising a group of n receivers R1 . . . Rn which exchange information in order to recover the data content of all the bursts simultaneously. As shown in
The performance of the group of receivers is investigated below. Each user link is characterized by an individual Packet Error Rate (PER) counter. In every slot, there is a counter which measures how many user bursts have been received without errors (n users transmit and n′ are received without errors). The statistical mean of n′ is an indication of the receiver effectiveness assuming the type of multiple-access described above. With an ideal receiver n′=n for n=1, 2, . . . in practice there is a threshold beyond which n′<n and n′ approaches zero as n increases. The ratio n′/n is the effective throughput indicating the average number of successfully decoded bursts. Of course, when the bursts of n′ terminals are decoded there is an average PER associated with each terminal. Due to the unscheduled nature of return transmissions and the satellite long round trip, transmission coordination in terms of timing, frequency and power may not be possible. Therefore, another requirement is that the multi-burst system should operate without the need of accurate power control. Timing and frequency may be estimated accurately if the terminal is able to determine its location relative to the satellite, for example using GPS. However, the proposed system capacity may be increased if the transmissions are not employing accurate timing, frequency and power.
The bursts preferably fall within the time slot and frequency boundaries for avoiding adjacent channel interference and their power is preferably limited so that extensive co-channel interference is also avoided in nearby spot beams. The propagation channel parameters may contain random elements that are unique for each burst, which may improve the overall capacity. On the other hand, accurate estimation of small disturbances may make demodulation more difficult and performance loss may be experienced.
Performance Results
The results in this section are for 20 ms bursts with rate 1/3 turbo coding. The receiver throughput is presented in
With a single burst receiver, the maximum throughput is of course 1 burst. When two bursts arrive in a single slot processed by a conventional single burst receiver, about a quarter of them are detected correctly on average, since the fading occasionally produces great energy difference between the two bursts and the strongest one can be decoded correctly. On the other hand, using a multi-burst receiver the average maximum throughput per slot is 3 bursts. The average PER is equivalent to the burst error rate shown in
The receiver architecture may be a trade-off between complexity and performance aiming at a ‘reasonable’ complexity solution. Further results are presented assuming 80 ms bursts with rate 1/5 turbo coding and three types of receiver: The first type uses the UW for channel state estimation and demodulation (UW Receiver). This has moderate complexity and it is very similar to the conventional receiver. The second type uses a type of iterative receiver or Turbo receiver in which the demodulator and FEC decoder iterate exchanging information, as disclosed for example in J. Hamkins & D. Divsalar, “Coupled Receiver/Decoder for Low Rate Turbo Codes”, ISIT, Yokohama, Japan, 2003. This receiver provides near optimum performance at the expense of higher complexity which scales according to channel conditions. Finally, the third type employs a perfect demodulator (Perfect Receiver). This is not practically realizable but sets the benchmark against which the implementation loss of the other two types mentioned above can be measured.
The receiver throughput is presented in
It should be noted that the available Eb/No is extensive and well above the fading margin required. However, this is an example of how one can trade power for capacity. Using a multi-burst receiver the average throughput is shown in
Controlled Interference
In the TCDMA embodiment described above, variations in power, time and frequency between the bursts transmitted by different terminals assist the receiver in discriminating between and successfully decoding the different bursts. Hence, in contrast with conventional methods, system performance may be improved by frequency and timing errors. In another embodiment, this effect is enhanced by the transmitting terminals introducing a deliberate random waveform variation, such as a power variation δP, timing variation δt and/or frequency variation δf for each burst.
In one example, a terminal applies a random or pseudo-random timing variation of up to a predetermined fraction, such as 5-10%, of the burst duration. In another example, a terminal applies a random or pseudo-random frequency variation of up to a predetermined fraction, such as 10%-20%, of the burst bandwidth.
Power uncertainty, caused for example by fading channel and random path loss, may also assist in discrimination between bursts. The power uncertainty may be enhanced by a transmitting terminal applying a random power variation to its bursts, with the maximum variation being varied by the system and communicated to the terminals using a signalling protocol in a similar fashion to the maximum timing and/or frequency variation described above with reference to
Where one burst is received with significantly higher power than other, interfering bursts, it may not be necessary to use a multi-user detection technique to decode the higher power burst; instead, this burst may be decoded using a simple single user decoder, with the other bursts being treated as noise. The decoded burst may then be subtracted from the received signal, and the remaining bursts may then be decoded by multi-user detection. In one embodiment, the high power bursts are transmitted to a conventional receiver, for example in a conventional system, and the low-power bursts are transmitted to an independent multi-user system that shares channels with the conventional system. In that case, the multi-user system does not need to decode the high power bursts, but merely identifies their characteristics sufficiently to be able to cancel them from the received signal prior to multi-user decoding.
The burst duration may be substantially equal to the time slot duration, unlike the embodiment shown in
Transmitter Control Logic
If the terminal has been allocated specific time and frequency slots in the return link (step S2), the terminal does not wait a random number of slots but proceeds to generate the rate signalling SDU (S10), encode the burst using the specified coding rate (S11), which may have been specified by the system as part of the allocation, sets power, frequency and timing offsets allocated by the system (S12) and transmits the burst (S9).
Shared Random Access and Reserved Allocations
In embodiments of the invention, scheduled and unscheduled (random access) communications can share return slots. In particular, terminals are permitted to transmit unscheduled signalling bursts, such as reservation requests, in time and frequency slots that are allocated to scheduled data communications. This allows much more efficient use of bandwidth compared to conventional systems in which channels, such as time slots and/or frequencies, are dedicated to unscheduled traffic. Sufficient bandwidth is conventionally allocated to such dedicated channels so as to maintain the probability of collisions below a certain level; hence, such conventional dedicated channels make very inefficient use of bandwidth. In contrast, in a system according to an embodiment of the present invention, there is little or no need for such dedicated channels; instead, unscheduled traffic shares bandwidth with scheduled traffic.
Burst Design for Reserved Allocations
A terminal may be allocated a plurality of sequential slots for continuous transmissions. In this case, the burst design may be as shown in
A similar burst structure may also be applied to unreserved bursts, with similar advantages at the expense of reduced bandwidth efficiency.
As an alternative to the distributed UW, pilot symbols may be distributed periodically into the bursts; pilot symbols are typically not unique to each transmitter and are therefore less able to assist with discrimination between bursts, but nevertheless assist in tracking the channel state.
In a continuous transmission, a transmitter may re-order the bursts prior to transmission such that data bits and related parity bits are spaced apart by more than one burst; in other words, they are not in adjacent bursts. This may reduce the sensitivity of the transmission to burst noise, but increases the latency of transmissions.
Multiple Bursts per Transmitter
In the above embodiments, it has been assumed that each of a plurality of bursts that occupy the same time slot is from a different transmitter, but this is not essential; the same transmitter may transmit multiple bursts in the same time slot if the bursts have sufficiently different waveform or propagation characteristics. In the former case, the transmitter may apply a frequency, timing and/or power offset between the bursts; this offset need not be random, but may be selected to ensure a sufficient offset. In the latter case, the transmitter may have a plurality of antennas exhibiting spatial, polarisation or other diversity so that the propagation characteristics from each antenna can be discriminated at the receiver, and may transmit a different burst from each antenna. The multiple bursts may each have a different data content, so that the transmission rate of the transmitter is increased.
Alternative Embodiments
Although at least some of the above embodiments have been described with reference to a satellite communications system, aspects of the invention are applicable to other types of wireless communication system, such as terrestrial communications systems. The embodiments may be applied to transmission and reception on a return link i.e. to transmissions from wireless terminals to a central receiver, but are more generally applicable to wireless multiple access systems. It is presently preferred that the receiver architecture be centralised due to the complexity of the receiver, but it is likely that such complexity could be introduced into user terminals in the near future, with predicted increases in processor power and reductions in processor cost and size. In that case, aspects of the invention could be applied to decentralised multiple access systems such as Aloha, wireless mesh networks and the like.
Although turbo codes are used in the preferred embodiments, other FEC (forward error correction) codes may be used. Parallel concatenated turbo codes are currently preferred because of their near-ideal coding performance, their ability to perform at very low coding rates, and their applicability to soft-in/soft-out decoding. Other high-performance codes such as serial concatenated turbo codes and low-density parity codes (LDPC) do not satisfy all of these requirements; however, were other codes to be discovered that satisfy these requirements, they would also be applicable in embodiments of the present invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| 0823132.6 | Dec 2008 | GB | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/GB2009/002377 | 10/5/2009 | WO | 00 | 10/17/2011 |
