GB2398205A - Subtraction of interfering signals in a relay radio network - Google Patents

Abstract

Signals which have been previously relayed by a repeater station are subtracted, so that the transmission bandwidth may be re-used. The transmitting and receiving stations may be provided at randomly distributed locations and a chain may be formed of repeated signals. By subtracting the signal now being transmitted by subsequent repeater stations in the chain, interference is reduced. The relay chain may use two alternating frequency channels to transmit the signal, and may delay the repetition until a later time. The cancellation may be performed by differencing means of the signal previously relayed, delayed and passed through a linear filter. The filter minimises the output power of the differencing means or the correlation of the output and the earlier relayed transmission, and may be adjusted recursively.

Description

"Radio Relaying with Improved Spectrum Efficiency in Distributed Radio

Telecommunication Networks"

DESCRIPTION

This invention relates to networks of radio stations in which information is successively relayed from one to the next over a number of hops. Examples of such networks are microwave point to point relay systems for transmitting telecommunications signals over long distances and also in distributed or meshed lo networks in which a plurality of such transmitting and receiving stations are provided at randomly distributed locations and in which switching circuitry is provided within the stations themselves for routing of calls between stations in the network utilising other stations in the network for relaying of such calls where necessary. It specifically addresses a novel technique to improve the spectrum efficiency of such networks by enabling the successive links in a relay chain of such stations to operate on the same frequency at the same time, by means of interference cancellation techniques.

In many countries, although there may be a telephone service to towns and some principal villages, the majority of the population has no effective access to telephones. There is a need in such countries for a network of telephones at such a density that substantially the whole of the population lives no more than a few kilometres from a public telephone. However this would require installation of a network comprising a large number of widely spaced telephones which would be prohibitively expensive if a conventional wired telephone system is used. The paper "A Distributed Rural Radio System for Developing Countries", S.A.G. Chandler, S.J.

Braithwaite, H.R. Mgombelo et al., Fourth IKE Conference on Telecommunications, IKE Conference Publication No. 371, April 1993 describes a rural radio telephony system which, by virtue of its exchangeless network structure, is ideally suited to providing a basic telephone service to widely separated sites.

Such a radio telephony system uses a network of cooperating radio nodes which do not require a central exchange or interconnecting infrastructure. Each node consists of a transmitting and receiving station comprising digital transceivers, at least one telephone interface and controllers containing software implementing a protocol to effect the required communication control. Each transmitting and receiving station comprises a solar powered digital radio unit with one or more telephones, computers or other devices connected to it. Calls within a reasonable range (50 kilometres or so in reasonably favourable terrain) are made by direct station-to-station communication.

Beyond this range, however, calls must be relayed by other stations within the 0 network. Calls outside the area served by the network, or requiring an excessive number of relay hops, may be routed through gateway nodes into the public service telephone network.

International Published Patent Application No. WO 97/13333 discloses such a radio telephony Reference is also made in to the paper "Analysis and Simulation of a Distributed Rural Radiotelephone Network", S.A.G. Chandler and J. Ni, Fourth European Conference on Radio Relay System, 1114 October 1993, Conference Publication No. 386.

The normal requirement for a multi-hop relay chain is for each station transmitting to use its own frequency or timeslot, re-use of frequencies being avoided unless the strength of interfering signals are more than some margin below that of the desired signal. This obviously increases the demands on the spectrum allocation for the network. The same problem arises in a slightly different guise should code division multiplex be used rather than time division or frequency division multiplex.

In microwave point to point systems the use of highly directional antennae with high back to front ratios are, together with a change in polarization on successive hops, provide the solution. However the fact that the antennae for distributed networks such as described are usually omni-directional makes this impossible. The invention described herein is a means to enable successive relay stages to simultaneously use the same transmission bandwidth by subtraction of interfering signals received at station S that have been previously relayed by S. but now being relayed by a subsequent station.

The technique is not quite as simple as might first appear as interfering signals from subsequent relay stations may arrive at S at any phase and are likely to suffer distortion due to multipath or the phenomenon of the signal arriving via a number of distinct paths with different delays. However these effects may be modelled by a complex linear transversal filter whose coefficients may be determined by the correlation of the received signal with the transmitted signal, known by S as it has lo previously relayed this signal. A more effective variation on performing this process of subtraction in an open loop manner, though, is to adapt the coefficients according to the correlation of the signal after subtraction with the original transmitted signal.

This process is in fact almost identical to that used for echo cancellation in a number of applications. This is a well proven technique giving reliable results except in the case of strongly auto-correlated data sequences. Such sequences are generally avoided by scrambling or other means for other reasons any way.

There is an obvious limitation on the technique in that interference from relay stages prior to transmission to station S. as the technique is based on S knowing the interfering signal data sequence. However the majority of benefits to be gained from the technique can be obtained without violating causality.

In order that the invention may be more fully understood, embodiments of the invention will now be described, by way of example, with reference to the accompanying drawings, in which: Figure I is an explanatory diagram of a distributed circuit switched telecommunication network; Figure 2 is a block diagram of a transmitting and receiving station in such a network; Figure 3 is a coverage map showing the potential problem to be addressed.

Figure 4 is a block diagram of the signal processing comprising the invention; s Figure 5 is a timing analysis diagram showing the preferred time division duplex embodiment.

Figure l is a diagram showing the location of nodes in a hypothetical lo distributed circuit switched radio telecommunication network comprises a series of randomly located fixed nodes l at which transmitting and receiving stations are located. In addition to the network nodes 1, several gateway nodes 2 are shown providing call access to the public service telephone network 3 in which telecommunication takes places in conventional manner by way of wired links under centralised exchange control. As indicated by broken lines 4 in the figure, calls may be made between network nodes 1, or between a network node 1 and a gateway node 2, either directly when the nodes are close enough to one another, or by way of other nodes 1 which serve to relay the calls.

Figure 2 is a block diagram of a transmitting and receiving station 6 comprising at least one transmit/receive aerials 7, 8, at least one digital transceiver 9, lO, a number, which could exceptionally be zero, of interfaces ll, 12 to telephones, computers or other devices, and associated devices 13, 14, an interconnection means and a control unit 16 for effecting control of communication between stations within the network. Each station 6 can be used to terminate using the telephones or other devices 13, 14, as well as to relay calls by simultaneously using the two transceivers 9, 10, to receive and re-transmit the call information in the two directions.

Transceivers 9, 10 typically use separate frequency channels, but may also use time division multiplex to carry or relay calls using the same transceiver. s

Figure 3 illustrates the problem to be overcome with four hypothetical stations. 20,21,22 and 23 in a relay chain in which 20 sends a signal to 23. As radio stations cannot simultaneously transmit and receive signals on the same frequency, a relaying station must either use frequency or time division relaying. In the former, signals received on one frequency are re-transmitted on another, and in the latter case signals received in one timeslot are re-transmitted during the next. Figure 3 may be taken to represent the situation in one of the timeslots or frequencies in which 20 is transmitting a signal to 21 while 22 transmits a signal, previously received from 21, to 23. 24 and 25 represent the areas over which signals may be reliably received from 20 lo and 22 respectively. 26 and 27 represent the areas over which signals from 20 and 22 cause interference with the reception of other signals. The ratio of these areas depends on the form of modulation used, as well as other things. It can be seen that 21 will receive a signal from 22, though, of similar strength to that from 20, which will therefore prevent proper reception of the signal from 20. The problem will not be so bad if directional antennae are used, as might be the case with fixed point to point relaying, but even here the problem may still be significant, particularly due to the effects of multi-path reflections. However with dynamically routed relaying through a multiplicity of stations, this would appear to completely prohibit such frequency re- use without the use of steerable directional antennae.

Figure 4 shows a block diagram of an embodiment of the method of overcoming this problem according to the invention herein described. The signal from the antenna which, at 21, consists of the sum of signals from 20 and 22,, after amplification, filtering and frequency conversion, is applied as a signal 37 to a differencing means 30, which subtracts an estimate of the signal received from 22.

This is then processed by channel equalization in an equaliser 31, filtering (not shown for simplicity) and demodulation in a demodulator 32, in the normal way to produce an estimate 38 of the message transmitted by 20. This message is passed through a I delay 33 by an amount corresponding to the delay in relaying, which will greatly exceed the total delay spread due to filtering and multipath. The waveform is then regenerated by a modulation means 34 to produce the signal transmitted by 22. A transversal filter 35 emulates linear distortion of this signal caused by filtering, multipath, timing errors and phase errors etc. The coefficients of 35 are adjusted dynamically using the output of correlator 36 so as to minimise the correlation between the estimate of the signal transmitted by 20 and that of the signal transmitted by 22. It is worthy of note hat the coefficients of the channel equaliser can give a good first estimate of the coefficients of the transversal filter for signals being relayed in the reverse direction.

The method of adjustment of the filter described is almost identical to that lo used for the adjustment of echo cancellers, and the same considerations apply. In particular, it is important that the signals show good autocorrelation properties or false nulls may result. There are numerous refinement to the method described in the literature, but these are not pertinent to the basic idea.

It is worth observing that it would be possible to extend the cancellation process to interference from a further stage of relaying should that be required, but not from a previous stage, as for instance interference from 20 on the reception of the signal from 22 by 23. This is because of causality considerations. However in both cases, the interfering station is three hops away and its signal is likely to have been sufficiently attenuated not to cause any problems unless a very high order modulation scheme, like 1024 QAM were used.

Figure 5 shows the preferred embodiment using time division duplex/ time division relaying. Shaded blocks 51 and 52 indicate the period a station is transmitting information going from left to right and in the opposite direction respectively. Solid arrows indicate the intended signal transmission whereas dashed arrows indicate the inevitable interference to be cancelled out. It should be observed that as the stations at either end of a relay chain do not themselves transmit at the time that relay stations would do, the problem addressed in this patent does not arise unless the relay chain consists of three or more hops. In the case of a two hop call, it would be possible for the relay station to simultaneously transmit the signals to be forwarded in each direction using the cancellation technique of this patent. However simultaneous reception of two terminating stations would not be possible without using the technique of CV-CCMA described in the paper by All, F.H., Chandler, S. A.G., and Soysa, S. "Complex Valued Receiver for Multiuser Collaborative Coding Schemes", s Electronic I,etters Vol 31 No 5 pp341-342, March 1995.

Note that a corrupt signal received as interference by station 21 from station 22 would be almost certain to cause an error in reception of the next data frame currently being received from station 20. This would not result in further errors though as the now corrupted data frame would now be correctly forwarded unless another error event were to occur. This would approximately double the error rate. This has little effect on the necessity for using error correction either by forward error correction, (FEC) or automatic repeat request, (ARQ) to ensure reliable transfer of data. If correction were to be performed on a hop-by-hop basis, rather than end to end, the cancellation process would not permit the use of ARQ, necessitating the use of forward error correction.

Figure 5 shows as single duplex pair involved in relaying. This is a specific case of the more general case of a multi-slot TDMA frame in which a plurality of stations operating on the same frequency channel, relay from one sending station to one receiving station for each timeslot within the relay frame. A more complex example is shown in figure 6a to 6f, in which signals are relayed from station 20 to 23 by transmissions in the first timeslots of the even transmit phase (Figure 6a) and odd transmit phase (Figure 6d), and from station 20 to 26 l by transmissions in the second timelots of each phase (Figures 6b and be) and from 23 to 26 in the third timeslots (figures 6c and 6f). It should be noted that the coefficients of the transversal filters are different for each timeslot. Intended transmissions are indicated by solid arrows, and interfering signals which are cancelled as described herein are indicated by dotted arrows.

Claims (8)

1. A radio station effecting the cancellation of interfering signals from subsequent stations in a chain of such radio stations relaying a signal from one s station to another using the same frequency at the same time by means of the fact that the interfering signal is the same as that already relayed by the receiving station.

2. A network of radio station according to claim I wherein relaying is effected by successive stations in a relay chain alternating their use of two frequency lo channels to transmit signals.

3. A network of stations according to claim 1 wherein relaying may be performed on a single frequency by stations storing received signals and re transmitting these to the next station at a later time.

4. A network of stations according to claim 3 wherein relaying operation is substantially as described in figure 5.

5. A station according to claim 1 wherein cancellation is performed by subtraction by a differencing means of the signal previously relayed having been delayed and passed through a linear filter.

6. A station according to claim 5 wherein the aforementioned filter is adjusted to minimise the output power of the differencing means or the correlation between the output of the differencing means and the previously relayed signal prior to or following filtering.

7. A station according to 6 wherein the adjustment of the filter is performed in a recursive manner.

8. A radio station substantially as described above and illustrated in the accompanying drawings.