WO1995022208A1

WO1995022208A1 - Decorrelating receiver for asynchronous cdma channels

Info

Publication number: WO1995022208A1
Application number: PCT/US1995/001482
Authority: WO
Inventors: Michael S. Feeney
Original assignee: Motorola Inc.
Priority date: 1994-02-14
Filing date: 1995-02-03
Publication date: 1995-08-17
Also published as: JPH08509343A; CA2160382A1

Abstract

A communication device (100) utilizes pilot symbols to decorrelate a plurality of CDMA signals without using power control. The received signals are filtered in a matched filter bank (103) and sampled in a sampler (104). A vector of the samples is stored in a memory (107) and the inverse cross correlation matrix is calculated by a computing device (105). The matrix is then multiplied by the energy estimates in a multiplier (108). The result is then subtracted from the vector in a summer (109). The result of the subtraction is multiplied by the inverse correlation matrix in another multiplier (110) in order to decode the signals in a decoder (112). Energy estimates of the plurality of CDMA signals are generated in an energy estimator (111) and given polarities based on the known polarity of the pilot symbols. These estimates are used in the subsequent multiplication of submatrices.

Description

DECORRELATING RECEIVER FOR ASYNCHRONOUS CDMA

CHANNELS

T^→Pr m aλ FtelH

This invention is generally related to communication devices and more particularly to CDMA communication devices.

Background

In personal communication systems and cellular communication systems, multiple mobile users communicate through a centralized base station by sharing a common channel. In such a system, a multiple access scheme must be employed to allow the users to share the channel so that all users can reliably cominunicate. Traditionally, frequency division multiple access (FDMA) and time division multiple access (TDMA) have been used. Recently, direct sequence code division multiple access (DS-CDMA) has received much attention as an attractive alternative to FDMA and TDMA for use in urban digital mobile radio networks. In DS-CDMA, multiplexing is achieved by assigning a unique signature sequence to each mobile user; knowledge of a particular user's code is required at the base station in order to demodulate that user's date. One claimed advantage of CDMA, among others, is that it will support more simultaneous users than either FDMA or TDMA. A disadvantage traditionally associated with CDMA is the so called near-far problem, which can arise when the received energies of the mobile signals arrive at the base station with unequal strengths. Traditionally, power control has been suggested to overcome the near-far problem. However, power control adds a great deal of complexity to the system, and it does not eliminate the near-far effect, it only reduces it.

One solution to this problem is to decorrelate the incoming signals in their entireties using inverse decorrelation matrices . This technique is suggested by R. Lupas and S. Verdu in "Near-far resistance of multi¬ user detectors in asynchronous channels" published in IEEE Transaction on Communications, Volume 38, Number 4, pp. 496-508, April 1990. The decorrelating detector presented in this reference is known to completely remove the multiple-access-interference (MAI) from the k™¹ user's final decision statistic in a direct sequence code division multiple access communication system with K users without knowledge of the received energies, at the expense of enhancing the background additive white Gaussian noise. As originally presented, however, the decorrelating detector requires the inversion of an MK by MK correlation matrix, where M is the common message length of all users. For large M, the computation of the matrix inverse will not be practical. As an alternative implementation, it was shown in this reference that for the limiting case

M > oo, the decorrelating detector becomes a K-input K-output linear time-invariant filter with a transfer function determined by the cross- correlations between the users' signature sequences (spreading codes). This solution is not comprehensive, however, because it will not accommodate the addition and/or deletion of users to the system. Another approach suggests a time windowed decorrelating receiver. This appears in an article by S. S. H. Wijayasuriya, J. P. Mcgeehan and G. H. Norton titled "Sliding window decorrelating algorithm for DS-CDMA receivers", printed in Electronics Letters, Volume 28, Number 17, pp. 1596-1598, August 13, 1992. This approach suggests a rather complex scheme to estimate bit polarities by incorporating a rate 1/2 convolutional code into the transmitter chain of each of the users. This approach greatly restricts the overall application of such a scheme in a receiver. In addition, this approach requires interleaving/de interleaving of the incoming signals. These two requirements adds to the complexity of both the transmitters and receivers used in the system. Furthermore, the window suggested by this reference is a sliding one. A sliding window is one that slides the bits from one window to another. In other words, for a window length of 5 bits, the first window would contain bits 1 through 5 and the second window would contain bits 2 through 6 and so on. This scheme of windowing the incoming signals results in a large number of partially overlapping windows. Since each window must be multiplied by the inverse correlation matrix, this scheme will require great computational efforts. It can therefore be appreciated that a need exists for a receiver that would decorrelate incoming signals without the complexities of the prior art. Brief Description nf *HP Dra i gs

FIG. 1 shows a block diagram of a communication device in accordance with the present invention.

FIG. 2 shows a flow chart of the operation of a communication device in accordance with the present invention.

FIG. 3 shows a communication system in accordance with the present invention.

Detailed Description of the Preferr I Embodiment

In CDMA applications a receiver needs to decorrelate incoming signals which may have various power levels. The available techniques require huge computational efforts which result in reduced efficiency and increased delays. The present invention provides for a pilot symbol to be periodically added to the signal at the transmitter. At the receiver, these pilot symbols are used as separators in forming windows. Since the polarity of the pilot symbols are known, no estimation is required hence the windows don't have to overlap. Interference within the window are removed using inverse correlation matrix. Interference from outside of the window is minimized using the known polarities of the pilot symbol bits. This approach significantly reduces the delay and the overall complexity of the receiver.

The principles of the present invention will be better understood by referring to a number of figures and more particularly to FIG. 3 where a communication system in accordance with the present invention is shown. The system 300 includes a communication device 100 which acts as the base station for the system 300. A plurahty of radios 302, 304, and 306 each having an antenna 303, 305, and 307, respectively are also included. The radios 302, 304, and 306 communicate with each other using Direct Sequence Code Division Multiple Access (DS-CDMA) signals modulated using Pilot Symbol Assisted Modulation (PSAM). As is known in the art, the CDMA signal includes a signal that has been modulated via a sequence code, often called a "signature code." All the coded signals are then communicated using the same channel. The demodulation is accomplished via decorrelation which essentially separates the signals based on their individual signature codes. In the preferred embodiment, the communication device 100 is aware of the signature codes used by each radio 302, 304, and 306. This is because the device 100 either assigned the code to the radios to begin with or acquired information on the code otherwise. In order to reduce the effects of near-far problem on signals received by each of the radios 302, 304, and 306, the communication device 100 must first decorrelate the signals and then re-transmit them, synchronously. The synchronous transmission of signals provides for a simple demodulation at the receiving unit independent of the near-far problem. The elements of the communication device 100 provide for a mechanism by which a number of asynchronous DS-CDMA-PSAM signals are decorrelated without needing power control or introducing significant delay. Referring to FIG. 1, a plurality of DS-CDMA-PSAM are received at an antenna 101. These signals are coupled to a delay device 102 and a time delay analyzer 115. The received signal passing through the antenna is the sum of the individual DS-CDMA-PSAM signals transmitted from each of the K radios 302, 304, and 306, plus additive channel noise. The radio transmitters insert a pilot symbol: (a "+1", for example) into their outgoing data streams after every L^h data symbol. These pilot symbols are used to remove interference from the decorrelated signals.

The time delay analyzer determines the time delays associated with each of the received signals. These delays are sent to a bank of matched filters 103 and to a computing device 105. The delay device 102 delays the received signals until the matched filter bank 103 has obtained the time delays from the time delay analyzer 115. The matched filter bank 103 includes K number of filters each of which is matched to one of the code sequences of the incoming DS-CDMA-PSAM signals. The K outputs of the matched filters are coupled to a sampler 104. The sampler 104 samples each of its K input signals at the end of every symbol interval and discards every (L +1)™¹ set of inputs (pilot symbols). The set of K outputs is sent to a memory unit 107. The memory unit 107 stores L consecutive sets of K inputs, thus forming a vector containing L x K elements. When the sampler 104 discards the incoming set of pilot symbols, memory unit 107 will send its contents to a summer 109, thus emptying itself. The computing device 105 computes an inverse correlation matrix (or an inverse cross correlation matrix) from the timing delays sent from the time delay analyzer 115 and the known CDMA code sequences (signature sequences). This inverse correlation matrix is sent to a second 5 memory device 106. The memory device 106 sends part of the inverse correlation matrix (a submatrix) to a multiplier 108 and sends the entire matrix to a second multiplier 110. In the preferred embodiment where the pilot symbols are chosen to be of the same polarity, the multiplier 108 multiplies the submatrix from the memory unit 106 with the energy

10 estimates sent from an energy estimator 111. The result is sent to the summer 109. It is noted that when pilot symbols with different polarities are used the output from the energy estimator 111 must include information on the particular polarities of the pilot symbols. This may be achieved by incorporating the polarity of the pilot symbol in the energy

15 estimate via a mixer 114. A PSAM polarity memory unit 113 stores the polarity of each radio's pilot symbols. In general, some radios could choose pilot symbols (-1) and other could choose (+1). The mixer 114 multiplies the energy estimates by their corresponding polarities arrived from the memory 113. The output of the mixer 108 is coupled to the

20 summer 109.

The summer 109 subtracts the output of the mixer 108 from the vector sent by the memory unit 107. This step removes interference from symbols outside the current data window. The mixer 110 multiplies the inverse correlation matrix with the vector free of outside interference.

!_-5 This "decorrelates" the symbols in the current data window and enables the communication device 100 to make symbol polarity decisions independently of the received energy levels (i.e., immune to the near-far effect). The output of mixer 110 is sent to the energy estimator 111 and to the decoder 112. The energy estimator 111 estimates the received energy

30 levels of the pilot symbols (outside interferences). The results are sent to the mixer 108 for use with the next data window. For the very first data window, the energy estimator 111 produces energies equal to zero. The decoder 112 forms the final symbols polarity decision for all symbols of each radio within the current data window. The output of the decoder 112

35 is sent to a synchronous transmitter 113 for synchronous transmission to the radios. Referring to FIG. 2 a flow chart of the operation of the communication device 100 in accordance with the present invention is shown. From a start block 200 a plurality of DS-CDMA-PSAM signals are received. These signals are delayed, block 206, and then filtered via the 5 bank of matched filters 103, block 208. The filtered signals are then sampled, block 210 before being stored in the memory 107 as a vector consisting of L consecutive sets of samples, block 212. Next, the inverse correlation matrix and its submatrices are computed, block 214. The multiplier 110 multiplies submatrices by the polarized estimates of

10 received energies during the previous window, block 216. It is noted that the first energy estimates are zero as there is no previous window from which estimates may be generated. At block 218, the summer 109 subtracts resulting interference from the vector of matched filter outputs to produce a modified vector. The modified vector is multiplied with the

15 inverse correlation matrix, block 220. Before the plurality of signals are decoded, block 222, the energy estimates are formulated, block 221. Once the signals have been decoded they are transmitted synchronously to the radios 302, 304, and 306, as desired.

In summary, the device 100 removes interference in a CDMA

20 system without requiring power control by passing the received signal (composite waveform of all the radios 302, 304, and 306) through a bank of matched filters, one filter matched to each allowed signature sequence. The outputs of the matched filters are sampled appropriately and several consecutive outputs (forming a window) are stored in the memory 107.

-2> The device 100 computes an inverse correlation matrix from the known active signature sequences and the acquired relative time delays between the radios' received signals. The device 100 then multiplies this matrix by the vector of stored matched filter outputs, in effect "canceling out" the interference or "decorrelating" the signals. Before this multiplication can

30 take place, the device 100 must first subtract out an estimate of the received energies including their polarities of all the radios 302, 304, and 306 immediately before and after the current window. The polarities of the radio signals before and after the window are known since pilot symbols are transmitted between every window. The received energy estimates are

35 obtained by taking the magnitudes of the previous window's decorrelating solution. Finally, after multiplying by the inverse correlation matrix, the receiver forms its final polarity decisions (+1 or -1) based upon the signs of the resulting outputs. By decorrelating the radios' signals, the device 100 is eliminating any dependence of one radio's signal upon the received energy of another radio's signal.

To illustrate the way in which the device 100 works, suppose we have a two user system and the decorrelating solution for the first window produces the resulting vector [3.561, -9.348]T where T denotes the transpose operation. Then the device 100 would decide that the first bit of user 1 was a +1 and the first bit of use 2 was a -1. The receiver would then use 3.561 as the energy estimate of the first user, and 9.348 as the energy estimate of the second user. These estimates would be subtracted off of the next window's values for user 1 and user 2 before multiplying again by the inverse correlation matrix. This process then continues until the signals are decorrelated.

A significant benefit of the present invention is that by using only a small portion of the transmitted sequences at a time (a window), the decoding delay is kept small, the storage requirements are kept small, and a much smaller matrix needs inversion. This saves in the complexity of the device 100 and produces decorrelated results faster with much higher efficiency. In addition, incorporating pilot-symbol-assisted- modulation (PSAM) into the system, the device 100 does not need to form estimates of the polarities of the bits preceding and following each window. Between every window, all the radios 302, 304, and 306 transmit a pilot bit of known polarity. Thus the device 100 knows the polarity of all the bits between the windows and no estimation scheme is needed. In addition, the windows suggested here separate the data into non- overlapping (non-sliding) blocks. For instance, if a window length of 5 bits is chosen, then the first window would contain bits 1 through 5, the second window would contain bits 6 through 10, and so on. This is much more efficient than the windowing scheme proposed in the prior art which utilizes sliding windows.

What is claimed is:

Claims

1. A communication device, comprising: a receiver for receiving a plurality of DS-CDMA-PSAM signals, the receiver comprising: a delay circuit for establishing delays associated with the plurality of DS-CDMA-PSAM signals; a bank of matched filters for filtering the plurality of DS- CDMA-PSAM signals to produce filtered signals; a sampler for sampling the filtered signals to produce sampled signals; a memory for storing a plurality of sampled signals preceded and followed by pilot symbols in order to form a window of samples; a computing device for computing the correlation matrix of the plurality of DS-CDMA-PSAM signals to provide an inverse correlation matrix; means for removing from the window of samples the effect of samples that precede and follow the window of samples in order to provide a vector free of outside interference;

^• a multiplier for multiplying the vector with the inverse correlation matrix in order to decorrelate the plurality of DS-CDMA-PSAM signals to produce decorrelated signals; and a transmitter for transmitting the decorrelated signals synchronously hence minimizing cross interference.

2. The communication device of claim 1, wherein the means for removing includes: an energy estimator for estimating received energy levels of signals outside the window of samples and adding polarity to the energy levels using the pilot symbols preceding and following the window of samples in order to produce polarized energy estimates of preceding and following samples; and a summer for subtracting the polarized energy estimates from the window of samples in order to provide a vector free of outside interference.

3. The communication device of claim 1, wherein the plurality of DS- CDMA-PSAM signals include a signal having a uniform pilot symbol.

4. The communication device of claim 3, wherein the uniform pilot symbol includes a +1.

5. A communication device, comprising: a receiver for receiving a plurality of DS-CDMA-PSAM signals, the receiver comprising: a delay circuit for establishing delays associated with the plurality of DS-CDMA-PSAM signals; a bank of matched filters for filtering the plurality of DS- CDMA-PSAM signals to produce filtered signals; a sampler for sampling the filtered signals to produce sampled signals; a memory for storing a plurality of sampled signals preceded and followed by a pilot symbol in order to form a window of samples; a computing device for computing the correlation matrix of the plurality of DS-CDMA-PSAM signals to provide a inverse correlation matrix; an energy estimator for estimating received energy levels of signals outside the window of samples and adding polarity to the energy levels using the pilot symbols preceding and following the window of samples in order to produce polarized energy estimates of preceding and following samples; a summer for subtracting the polarized energy estimates from the window of samples in order to provide a vector free of outside interference; a multiplier for multiplying the vector with the inverse correlation matrix in order to cancel interference from outside the window of samples and decorrelate the plurality of DS-CDMA-PSAM signals to produce decorrelated signals; and a transmitter for transmitting the decorrelated signals synchronously hence minimizing cross interference.

6. A receiver for receiving a plurahty of DS-CDMA-PSAM signals, the receiver comprising: a delay circuit for estabHshing delays associated with the plurality of DS-CDMA-PSAM signals; a bank of matched filters for filtering the plurality of DS-CDMA- PSAM signals to produce filtered signals; a sampler for sampling the filtered signals to produce sampled signals; a memory for storing a plurality of sampled signals preceded and followed by a pilot symbol in order to form a window of samples; a computing device for computing the correlation matrix of the plurality of DS-CDMA-PSAM signals to provide an inverse correlation matrix; means for removing from the window of samples the effect of samples that precede and follow the window of samples in order to provide a vector free of outside interference; a multiplier for multiplying the vector with the inverse correlation matrix in order to cancel interference from outside the window of samples and decorrelate the plurality of DS- CDMA-PSAM signals to produce decorrelated signals; and a decoder for decoding the decorrelated signals to produce decoded signals.

7. The receiver of claim 6, fuxther including a digital to analog converter to convert segments of the decoded signals into analog signals.

8. A method of receiving and decorrelating a plurality of DS-CDMA signal, comprising the steps of: receiving a plurality of DS-CDMA-PSAM signals; estabHshing delays associated with the plurality of DS-CDMA-

PSAM signals; filtering the plurality of DS-CDMA-PSAM signals to produce filtered signals; sampling the filtered signals to produce sampled signals; storing a plurality of sampled signals preceded and followed by a pilot symbol in order to form a window of samples; computing the correlation matrix of the plurality of DS-CDMA- PSAM signals to provide an inverse correlation matrix; removing from the window of samples the effect of samples that precede and follow the window of samples in order to provide a vector free of outside interference; and multiplying the vector with the inverse correlation matrix in order to cancel interference from outside the window of samples and decorrelate the plurality of DS-CDMA-PSAM signals to produce decorrelated signals.

9. The method of claim 8, wherein the step of removing comprises the steps of: estimating received energy levels of signals outside the window of samples and adding polarity to the energy levels using the pilot symbols preceding and following the window of samples in order to produce polarized energy estimates of preceding and foUowing samples; subtracting the polarized energy estimates from the window of samples in order to provide a vector free of outside interference; and

10. A communication system, comprising: a plurality of radios; a base station for controlling the communication between the plurality of radios, the base station comprising: a receiver for receiving a plurality of DS-CDMA-PSAM signals from the plurality of radios, the receiver comprising: a delay circuit for estabHshing delays associated with the plurality of DS-CDMA-PSAM signals; a bank of matched filters for filtering the plurality of DS-CDMA-PSAM signals to produce filtered signals; a sampler for samphng the filtered signals to produce sampled signals; a memory for storing a plurality of sampled signals preceded and followed by pilot symbols in order to form a window of samples; a computing device for computing the correlation matrix of the plurality of DS-CDMA-PSAM signals to provide an inverse correlation matrix; means for removing from the window of samples the effect of samples that precede and follow the window of samples in order to provide a vector free of outside interference; a multipHer for multiplying the vector with the inverse correlation matrix in order to decorrelate the pluraHty of DS-CDMA-PSAM signals to produce decorrelated signals; and a transmitter for transmitting the decorrelated signals synchronously to the plurality of radio hence minimizing cross interference.