WO2000045138A2

WO2000045138A2 - Dsss receiver

Info

Publication number: WO2000045138A2
Application number: PCT/IL2000/000047
Authority: WO
Inventors: David Hirshberg
Original assignee: State Of Israel - Ministry Of Defense
Priority date: 1999-01-28
Filing date: 2000-01-24
Publication date: 2000-08-03
Also published as: AU2316200A; IL128262A0; WO2000045138A3

Abstract

A receiver for a signal that includes several symbols modulated with a pseudonoise sequence and that travels to the receiver along a plurality of paths in a channel. The receiver includes a matched filter (34) for correlating the signal with the pseudonoise sequence to produce a correlation function. The receiver also includes a delay estimator (36) for estimating a delay for each of the paths from the correlation function and a number of correlators (40) for correlating the signal with the pseudonoise sequence in accordance with the delays.

Description

DSSS RECEIVER

FIELD AND BACKGROUND OF THE INVENTION The present invention relates to spread spectrum communication and, more particularly, to a receiver for receiving bursty signals that travel along multiple propagation paths.

The principles of direct sequence spread spectrum (DSSS) communication are well-known. The message to be sent is modulated by a pseudonoise (PN) sequence. The received message is demodulated ("despread") by correlating the received signal with the same pseudonoise sequence.

The message consists of a string of symbols. All the symbols are of equal duration T_s. The modulation step consists of multiplying each symbol by the pseudonoise sequence to produce a sequence of "chips" which also are of equal duration T_c; but if there are N chips in the pseudonoise sequence then T =TJN. The despreading is performed by correlating the received sequence of chips with the pseudonoise sequence; but for this to be done correctly, the correlation must start at points in the received sequence of chips that correspond to the beginning of symbols. Therefore, the despreading includes two phases: acquisition, in which the receiver synchronizes itself with the incoming sequence of chips, to determine where in the incoming sequence of chips to start the correlation, and tracking, in which the receiver maintains synchronization while recovering the symbols by sequential correlation of the received sequence of chips.

Two forms of hardware are commonly used to effect the correlation: matched filters and correlators. Figure 1 is a block diagram of a matched filter 10. Matched filter 10 includes a tap delay line 12 for receiving and storing up to N chips, N multipliers 14 for multiplying each received chip by a different chip of a pseudonoise sequence 18, and N adders 16 for accumulating the resulting products. Every T_c time units, a new received chip enters tap delay line 12. The chips in tap delay line 12 are multiplied collectively by the chips of pseudonoise sequence 18, and, at any particular time, the output is the value of the correlation of the chip sequence with pseudonoise sequence 18 at that time. Figure 2 is a block diagram of a correlator 20. Correlator 20 includes one multiplier 22 and one adder 24. Every T_c time units, a different incoming chip is multiplied by a different chip of the pseudonoise sequence. The products are accumulated, and the sum is fed back as shown, so that after every N multiplies and adds, i.e., after NT_C time units, correlator 20 produces one value of the correlation of the chip sequence with the pseudonoise sequence.

Matched filter 10 has the advantage of speed over correlator 20. By performing N multiplies and adds to every one multiply and add of correlator 20, matched filter 10 produces N correlation values for every one correlation value produced by correlator 20. This is an advantage when messages are transmitted in bursts. Each burst consists of a preamble followed by the data of the message proper. Every time a new burst is received, acquisition must be initiated anew. A receiver based on matched filter 10 is capable of effecting acquisition and synchronization before the entire preamble is received, so that no data is lost. By the time a receiver based on correlator 20 has synchronized itself with an incoming signal, the transmission may be over. This disadvantage of correlator 20 often is mitigated to a certain extent by using banks of correlators 20, with each correlator in the bank receiving the incoming chips with a different delay.

Correlator 20 has two advantages over matched filter 10. First, receiver hardware based on matched filter 10 is considerably more extensive than receiver hardware based on correlator 20. Second, matched filter 10 is inherently limited to a pseudonoise sequence 18 whose duration is the same as the duration of one symbol.

In most implementations of a DSSS receiver, the representations of the received chips that are input to matched filter 10 or correlator 20 for despreading have in-phase and quadrature components. In a digital implementation of a DSSS receiver, the representations are complex numbers, each of which can be expressed as a real part (corresponding to the in-phase component) and an imaginary part (corresponding to the quadrature component), or equivalently, as an amplitude and a phase. Therefore, the representations of the symbols that are output by matched filter 10 or by correlator 20 also are complex numbers.

Figure 3 illustrates a situation in which a pseudonoise sequence longer than one symbol is useful. Consider a simple 3-bit message ("101") that has been modulated with a pseudonoise sequence to produce a signal for transmission. Figure 3A shows, for reference, the correlation function obtained by correlating this signal with the same pseudonoise sequence as was used for modulation. (Note that the zero bit of the message is represented by a negative peak.) If this signal is sent to a receiver via a channel that includes a single propagation path, this is the correlation function that the receiver produces. The receiver then recovers the message from the correlation function by thresholding the absolute values of the peaks of the correlation function and inspecting the signs of the peaks. If the signal is sent via a channel that includes two propagation paths, with the second propagation path delayed relative to the first by slightly more than T_s, then the receiver produces the correlation function shown in Figure 3B: the sum of the correlation function of Figure 3 A and a delayed copy of itself. The correct interpretation of this correlation function may be difficult. If the delay is within T_c of an integral multiple of T_s, then the correct interpretation of this correlation function requires the use of equalization techniques, which add to complexity and may result in significant degradation of performance.

This problem can be solved by using a composite pseudonoise sequence that is longer than the largest path-to-path propagation delay in the channel. In the particular example shown in Figure 3, the pseudonoise sequence should be twice as long as one symbol. The composite pseudonoise sequence is formed by concatenating two pseudonoise sequences, each of which is one symbol long, and which are approximately mutually orthogonal. The two one-symbol-long pseudonoise sequences then are subseqences of the composite pseudonoise sequence. In effect, the odd-numbered symbols are modulated with the first subsequence and are despread by correlation with the first subsequence; and the even-numbered symbols are modulated with the second subsequence and are despread by correlation with the second subsequence. When the first symbol arrives via the delayed propagation path, both it and the second symbol arriving via the first propagation path are despread by correlation with the second subsequence; but because the second subsequence is approximately orthogonal to the first subsequence, only a peak corresponding to the second symbol appears in the correlation function.

The correlation functions illustrated in Figure 3 are real. As noted above, in general the correlation functions produced by matched filter 10 or by correlator 20 are complex. The corresponding representations of the symbols are the values of the correlation functions at their peak absolute values. Ideally, in a channel with only one propagation path, binary signaling and coherent reception, the imaginary parts of the correlation functions are much smaller than their real parts, and the phases of the peaks are close to either zero (representing a one bit) or π (representing a zero bit). In practice, the different representations of a symbol that correspond to different propagation paths have different amplitudes and phases, and the phases can have any value in the interval (-π,π].

If the modulated signal transmitted into a multipath channel that includes L different propagation paths is s(t), then the signal received by the receiver is

- τ_/) , where / indexes the propagation paths, and each path is

characterized by the following parameters: an amplitude g a phase θ,- and a delay τ,-. j is the square root of -1, δ represents the Kronecker delta function and "*" represents convolution. For each transmitted symbol, the output of correlators 20 is a vector of L representations h_te ' , l=\,...JL, of that symbol. If/?_/ is properly normalized, and in the absence of noise and of interference other than that associated with multipath propagation, /? =g_/ and φ_/=θ +θ_m where θ_m is the phase modulation angle (e.g., 0 or π for BPSK modulation; 0, π/2, π or 3π/2 for QPSK modulation). If the largest path-to- path propagation delay exceeds T_s, then the representations of different symbols are interleaved in time and must be disentangled by the receiver. In order to minimize the bit error rate, representations of the same symbol must be combined into a single value. A receiver with this functionality is known in the literature as a RAKE receiver. See, for example, R. Price and P. E. Green, Jr., "A communication technique for multipath channels", Proc. IRE vol. 46 (March 1958) pp. 555-570, and J. G. Proakis, Digital Communication, third edition (McGraw-Hill, 1995) pp. 798- 806. Two classes of combining schemes are used:

(a) non-coherent schemes, which use only amplitude information; and

(b) coherent schemes, which use both amplitude information and phase information. The optimum coherent combination, also called the "maximum ratio combination", is as follows:

y = ∑h,g, exp[j(φ_l - #_/)]

/=! where g_; are estimates of the true amplitudes of the distortions due to the propagation paths, and θ_t are estimates of the true phases of the distortions due to the propagation paths. The estimates g_; are obtained by averaging in a several-symbols-long sliding window around the target representation. Obtaining the estimates 0, is more complicated because the phases depend on the sign of the associated symbol; but algorithms such as NINA (Andrew J. Niterbi and Audrey M. Niterbi, "Nonlinear estimation of PSK-modulated carrier phase with application to burst digital communication, IEEE Trans. Information Theory vol. IT-29 (July 1983), pp. 543-551) are well-known.

All phase estimation algorithms suffer from performance degradation in the presence of symbol modulation. This performance degradation may be significant in the case of a RAKE receiver, in which the noise-to-signal ratios associated with separate propagation paths are higher, and may be much higher, than the noise-to- signal ratio inherent in the total signal.

There is thus a widely recognized need for, and it would be highly advantageous to have, a spread spectrum receiver that combines fast acquisition with resistance to multipath interference.

SUMMARY OF THE INVENTION

According to the present invention there is provided a receiver for a signal that includes a plurality of symbols modulated with a pseudonoise sequence and that travels to the receiver along a plurality of paths in a channel, including: (a) a matched filter for correlating the signal with the pseudonoise sequence to produce a correlation function; (b) a delay estimator for estimating a delay for each of the paths from the correlation function; and (c) a plurality of correlators for correlating the signal with the pseudonoise sequence in accordance with the delays. According to the present invention there is provided a receiver for a signal that includes a plurality of symbols modulated with a pseudonoise sequence and that travels to the receiver along a plurality of paths in a channel, including: (a) a plurality of correlators; and (b) a controller for configuring the plurality of controllers to alternately: (i) correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function, and (ii) for each of the paths, correlate the signal with at least a portion of the pseudonoise sequence in accordance with a delay associated with the path.

According to the present invention there is provided, in a communications system in which a message including a plurality of symbols is modulated with a pseudonoise sequence and the resulting signal is transmitted along a certain number of propagation paths, a method of receiving the signal and recovering the symbols, including the steps of: (a) providing a receiver including: (i) a matched filter and (ii) a plurality of correlators; (b) acquiring the signal, using the matched filter; and (c) tracking the signal, using at least one of the plurality of correlators.

According to the present invention there is provided, in a communications system in which a message including a plurality of symbols is modulated with a pseudonoise sequence and the resulting signal is transmitted along a certain number of propagation paths, a method of receiving the signal and recovering the symbols, including the steps of: (a) providing a receiver including a plurality of correlators; (b) configuring the plurality of correlators to correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function; (c) acquiring the signal, using the correlation function; (d) configuring at least one of the correlators to track the signal; and (e) tracking the signal, using the at least one correlator. According to the present invention there is provided a method of transmitting and receiving a message including a plurality of symbols, all of the symbols having a common symbol duration, including the steps of: (a) providing a pseudonoise sequence; (b) modulating the symbols with the pseudonoise sequence to produce a signal; (c) providing a receiver including: (i) a matched filter and (ii) a plurality of correlators; (d) acquiring the signal using the matched filter; and (e) tracking the signal using at least one of the correlators. According to the present invention there is provided a method of transmitting and receiving a message including a plurality of symbols, all of the symbols having a common symbol duration, including the steps of: (a) providing a pseudonoise sequence; (b) providing a receiver including a plurality of correlators; (c) configuring the plurality of correlators to correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function; (d) acquiring the signal, using the correlation function; (e) configuring at least one of the correlators to track the signal; and (f) tracking the signal, using the at least one correlator.

According to the present invention there is provided, in a communications system wherein a message including a plurality of symbols is transmitted to a receiver along each of a plurality of propagation paths, the transmission along each propagation path being effected with a certain delay, the message including a preamble and a data sequence, the preamble terminating in a synchronization sequence, the receiver producing, for each symbol of the data sequence and for each symbol of the synchronization sequence, a plurality of representations of that symbol, each representation corresponding to a unique one of the propagation paths and being produced at a time corresponding to the delay of the corresponding propagation path: a method for identifying and grouping the representations of each symbol of the data sequence for further processing, including the steps of: (a) sequentially storing the representations, as the representations are produced, in a plurality of sequentially addressed locations; (b) identifying a first representation of a first symbol of the data sequence; (c) for each subsequent representation of the first symbol, identifying a corresponding address increment; and (d) for each symbol of the data sequence, retrieving the representations of that symbol together in accordance with the address increments.

According to the present invention there is provided, in a communications system wherein, for each of a plurality of symbols, a plurality of representations is obtained, each representation having an amplitude and a phase, a method of combining the representations of each symbol, including the steps of: for each symbol: (a) identifying, from among the representations, a reference representation; (b) for each other representation: (i) estimating a true difference between the phase of the each other representation and the phase of the reference representation, and (ii) adjusting the phase of the each other representation in accordance with the estimated true difference, thereby producing a phase-adjusted representation; and (c) summing the phase-adjusted representations with the reference representation.

The principle of the present invention is to use a matched filter for acquisition and a bank of correlators for tracking. The pseudonoise sequence used is a composite pseudonoise sequence whose duration is an integral multiple of the symbol duration and whose subsequences are one-symbol-long pseudonoise sequences that are approximately mutually orthogonal. The duration of the pseudonoise sequence is selected to exceed all anticipated multipath delays. Only the first subsequence is used for acquisition, in the matched filter. A delay estimator receives the correlation function produced by the matched filter and estimates from it the number and relative delays of the propagation paths. This number of correlators is used to perform the tracking correlations, with the pseudonoise inputs to the correlators delayed in accordance with the estimated relative delays. In other words, one correlator is assigned to each propagation path.

A glance at Figures 1 and 2 shows that the essential components of matched filter 10, i.e., multipliers 14 and adders 16, are identical to the essential components of correlator 20, i.e., multiplier 22 and adder 24. Therefore, in preferred embodiments of the receiver of the present invention, the same bank of correlators is used for both acquisition and tracking. The pseudonoise code is supplied to the correlators via delay units. A controller adjusts the delays as appropriate: integral multiples of T_c for acquisition; estimated relative path delays for tracking.

The outputs of the correlators are combined in a RAKE combiner to give a robust estimate of the incoming message. A further aspect of the present invention is an improved method of coherently combining the correlator outputs. Instead of seeking to remove the phase distortion imposed by the various propagation paths (absolute correction), the phases of all the symbol representations are shifted into alignment with the phase distortion of the propagation path of highest estimated true amplitude (relative correction). To facilitate this coherent combination, the symbol representations coming from the correlators are stored sequentially as they are produced by the correlators, and representations of the same symbol are retrieved together and forwarded to the RAKE combiner. Although the present invention is exemplified herein with digital implementations thereof, it will be understood that the scope of the present invention also includes analog implementations.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention is herein described, by way of example only, with reference to the accompanying drawings, wherein:

FIG. 1 (prior art) is a block diagram of a matched filter; FIG. 2 (prior art) is a block diagram of a correlator;

FIG. 3 shows an example of DSSS multipath interference associated with a one-symbol-long pseudonoise sequence;

FIG. 4 is a partial block diagram of a first embodiment of a receiver of the present invention; FIG. 5 is a block diagram of an alternate embodiment of a correlator bank;

FIG. 6 shows various functions useful in explaining the operation of the embodiment of FIG. 4;

FIG. 7 is a high level block diagram of the digital portion of a second embodiment of a receiver of the present invention FIG. 8 shows the structure of a burst;

FIG. 9 is a block diagram of a symbol synchronizer;

FIG. 10 illustrates the order of storage and retrieval in the symbol synchronizer of FIG. 9.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

The present invention is of a spread spectrum receiver which can be used for bursty communication via a multipath channel.

The principles and operation of spread spectrum communication according to the present invention may be better understood with reference to the drawings and the accompanying description.

Referring again to the drawings, Figure 4 is a partial block diagram of a first embodiment 30 of a receiver of the present invention. In receiver 30, a radio signal received via an antenna 31 is filtered, down-converted and digitized in an analog front end 32. Front end 32 outputs a sequence of pairs of numbers, each pair representing the in-phase and quadrature (real and imaginary) components of one received chip. The time interval between successive pairs of numbers is an integral fraction of T_c, typically TJ2 or TJA. These pairs of numbers are sent to a matched filter 34, which effects acquisition, and to a correlator bank 38, which effects tracking. The correlation function produced by matched filter 34 is sent to a delay estimator 36, which analyzes the correlation function to obtain the delays Xp Delay estimator 36 passes the delays along to correlator bank 38. Correlator bank 38 includes a pseudonoise sequence generator 42, a delay unit 44 and several correlators 40, of which three are shown. L correlators 40 are actually used, one for each propagation path. Delay unit 44 provides the pseudonoise sequence from pseudonoise sequence generator 42 to each active correlator 40 with a delay equal to the corresponding propagation delay x_t. In that way, the output of each active correlator 40 is synchronized with the arrival of symbols along the corresponding propagation path. The outputs of correlators 40 go to a channel estimator 41, which produces true amplitude estimates g, and true phase estimates θ_l and to a RAKE combiner 46, which uses g, and θ, to combine the outputs of correlators 40 coherently. The output of RAKE combiner 46 is detected by a symbol detector 48. The operation of symbol detector 48 is discussed below, in the context of a preferred embodiment of RAKE combiner 46.

Although front end 32 produces number pairs at intervals shorter than T_c, the input to any one of the correlators of matched filter 34, or to any one correlator 40, is at intervals of T_c. The reason for producing number pairs at fractional intervals is to enable other correlators of correlator bank 38 to be used in delay lock loops (DLLs). For each correlator 40, there is a delay lock loop for maintaining synchronization of delay unit 42 and correlator 40 with the incoming symbols. Each delay lock loop includes two other correlators, a first DLL correlator that receives number pairs from front end 32 at times TJ2 earlier than the number pairs received by correlator 40, and a second DLL correlator that receives the same number pairs as the first DLL correlator, but delayed by T_c. As is well known in the art, the difference of the outputs of the two correlators of the delay lock loop provides a signal to delay unit 44 that keeps delay unit 44 synchronized with the symbols arriving via the propagation path to which correlator 40 is assigned. For clarity, only the components of receiver 30 that are germane to the present invention, and not, for example, the DLL correlators, are illustrated in Figure 4. Those ordinarily skilled in the art will be able to supply the components that are missing from Figure 4 to produce a complete block diagram of a functional spread spectrum receiver.

Figure 5 illustrates an alternate embodiment 50 of a correlator block that has the operational capabilities of both matched filter 34 and correlator block 38. For clarity, only four correlators 52 are shown in Figure 5, but it is to be understood that correlator block 50 includes at least as many chips as are produced by multiplying one symbol by the pseudonoise sequence, to enable correlator block 50 to function as a matched filter. To each correlator 52 is assigned a delay unit 54. Delay units 54 are controlled by a controller 56. To operate correlator block 50 as a matched filter for acquisition, each delay unit 54 after the first delay unit 54 is set by a controller 56 to delay the pseudonoise sequence from a pseudonoise sequence generator 58 by T_c, so that successive correlators 52 receive the pseudonoise sequence with delays of 0, T_c, 2T_C, 3T_C, etc. To operate correlator block 50 in tracking mode, delay units 54 whose correlators 52 are assigned to propagation paths 1-2 through l=L are set by controller 56 to delay the pseudonoise sequence by differences between successive delays X . For each correlator 52 that is assigned to a propagation path, two other correlators 52 are assigned to a delay lock loop for that propagation path, and the corresponding delay units 54 are set to delays that are respectively Tc/2 less and Tc/2 more than the total delay of the first correlator 52 that is assigned to that propagation path. As noted above, the pseudonoise sequence used for modulation and for despreading is a composite pseudonoise sequence consisting of several pseudonoise subsequences, each subsequence as long, in time, as one symbol, and all the subsequences being approximately mutually orthogonal. In the embodiment of Figure 4, pseudonoise sequence generator 42 generates the full pseudonoise sequence, for tracking; and matched filter 34 includes a pseudonoise sequence generator that generates only the first subsequence, for acquisition. In the embodiment of Figure 5, pseudonoise sequence generator 58 generates only the first subsequence during acquisition and generates the full pseudonoise sequence during tracking. The arrow from controller 56 to pseudonoise sequence generator 58 symbolizes the fact that controller 56 causes pseudonoise sequence generator to generate only the first subsequence during acquisition but to generate the full pseudonoise sequence during tracking. Also shown in Figure 5 is a multiplexer 57 that receives the outputs of correlators 52 during acquisition and passes the combined output, which is a complete correlation function such as is output by matched filter 34, to delay estimator 36.

Figure 6 illustrates a fragment 60 of a message and how the corresponding signal is processed by receiver 30 during tracking. Fragment 60 includes the six symbols (single bits in this case) 1, 0, 1, 1, 1 and 0, each of duration T_s. For transmission, fragment 60 is modulated with a pseudonoise sequence 62 that is a concatenation of three approximately mutually orthogonal pseudonoise subsequences PN-A, PN-B and PN-C, each also having a duration T_s, so that pseudonoise sequence 62 has a total duration o 3T_s. This total duration of 3T_S is selected in anticipation of transmission of the message via a communication channel having several propagation paths with a maximum relative delay of up to 3T_S. The resulting signal travels to receiver 30 in a channel that has two propagation paths, one delayed relative to the other by TJ2. Matched filter 34 correlates the received signal with only subsequence PN-A. As a result, correlation function 64 produced by matched filter 34 includes only peaks corresponding to the first and fourth symbols of fragment 60. Peaks corresponding to the other symbols are absent because subsequences PN-B and PN-C, which are used to modulate those symbols, are approximately orthogonal to subsequence PN-A. For reference, correlation function 66 also is shown. Correlation function 66 would be produced by three matched filters, each one using a different subsequence of pseudonoise sequence 62. Delay estimator 36 analyzes a correlation function, from the preamble of the message, that resembles correlation function 64 and decides to use two correlators 40 for tracking. The first correlator 40 is provided by delay unit 44 with a synchronous copy of pseudonoise sequence 62 and produces, synchronously with the end of each symbol, an output 68 that corresponds to the corresponding synchronous peak of correlation function 66. The second correlator 40 is provided by delay unit 44 with a copy of pseudonoise sequence 62 that is delayed by TJ2 and produces, synchronously with the end of each symbol, an output 70 that corresponds to the corresponding synchronous peak of correlation function 66 with a delay of TJ2. RAKE combiner 46 combines outputs 68 and 70 coherently, so that output function 60 is produced by symbol detector 48.

Figure 7 is a high level block diagram of the digital portion 80 of a second embodiment of a receiver of the present invention. In Figure 7, the solid arrows represent the flow of data (representations of chips or symbols) and the dashed arrows represent the flow of channel parameters such as delays, phases and amplitudes. As in the embodiment of Figure 4, the representations of the incoming chips are received by a matched filter 82 and by a correlator bank 86. Matched filter 82 effects acquisition by correlating the pseudonoise sequence with the chips of the preamble, and passes the resulting correlation function to a delay estimator 84, which determines the number of propagation paths and the corresponding delays X_/. Three correlators of correlator bank 86 are assigned to each of the propagation paths. One of the three correlators is provided with a pseudonoise sequence that is advanced relative to the corresponding delay τ_; by TJ2. Another of the three correlators is provided with a pseudonoise sequence that is retarded relative to the corresponding delay X by TJ2. The outputs of these two correlators is sent to DLL 88, which keeps the third correlator synchronized with the symbols of the corresponding channel propagation path. The carrier frequency used by front end 32 may differ from the actual carrier frequency of the radio signal, for example if frequency lock is not achieved. This mismatch is reflected in a slow drift of the phases θ_/. A frequency estimator 92 detects this drift and passes it to a frequency correction unit 94, which adjusts the phases of the representations of the symbols accordingly. A delay unit 90 delays the data flow long enough for frequency estimator 92 to complete its computations. See, for example, S. Bellini, C. Molinari and G. Tartara, "Digital frequency estimation in burst mode QPSK transmission", IEEE Trans Communication, Vol. 38 (July 1990), pp. 959-961 for an implementation of frequency estimator 92.

A symbol synchronizer 96, which is described in more detail below, receives the phase-corrected symbol representations from frequency correction unit 94 and sorts the symbol representations so that all L symbol representation that correspond to the same symbol received via the L propagation paths are processed together by an amplitude estimator 98, a phase estimator 102 and a RAKE combiner 104. Amplitude estimator 98 and phase estimator 102 compute estimates g, and θ_t of the true path amplitudes and phases. A delay unit 100 delays the data flow while amplitude estimator 98 and phase estimator 102 execute their computations. RAKE combiner 104 receives amplitude estimates g_l from amplitude estimator 98 and phase estimates θ_t from phase estimator 102 and uses them to combine the symbol representations coherently, as discussed below. As in the embodiment of Figure 4, the output of RAKE combiner 104 is processed by a symbol detector 106 to resolve the actual symbol^'s. The operation of symbol synchronizer 96 may be understood with reference to

Figure 8, which shows the structure of a typical burst 110: a preamble 112 of 128 symbols followed by a data sequence 118 of 896 symbols. Preamble 112 itself consists of an acquisition sequence 114 of 64 symbols followed by a synchronization sequence 116 of 64 symbols. Matched filter 34 or 82 acquires burst 110 before all of acquisition sequence 114 has been received via the fastest propagation path. Therefore, representations of the symbols of both synchronization sequence 116 and data sequence 118, as transmitted via all L propagation paths, are output from correlator bank 38 or 86.

To understand the operation of symbol synchronizer 96, it also should be noted that, just as correlator bank 86 includes a separate correlator for each expected propagation path, so frequency correction unit 94 includes separate hardware for each expected propagation path. Just as one correlator of correlator bank 86 is assigned to each detected propagation path, so one section of; frequency correction unit 94 is assigned to each detected propagation path. Figure 9 is a block diagram of symbol synchronizer 96. The particular embodiment of symbol synchronizer 96 that is illustrated in Figure 9 handles symbol representation from up to four different propagation paths a, b, c and d. The symbol representations go to a random access memory (RAM) 126, and also to a controller 124 via matched filters 120 and threshold detectors 122. One matched filter 120 and one threshold detector 122 is assigned to each propagation path: matched filter 120a and threshold detector 122a is assigned to propagation path a; matched filter 120b and threshold detector 122b is assigned to propagation path b; matched filter 120c and threshold detector 122c is assigned to propagation path c; and matched filter 120d and threshold detector 122d is assigned to propagation path d. The flow of symbol representations is represented in Figure 9 by solid arrows. The flow of control signals is represented in Figure 9 by dashed arrows.

Unlike matched filters 34 and 82, which are chip-oriented, matched filters 120 are symbol-oriented. Specifically, each of matched filters 120 produces its peak output only upon the receipt of all the symbols of synchronization sequence 116. Each threshold detector 122 notifies controller 124 of the receipt of output from the corresponding matches filter 120 that is in excess of a preset threshold. The receipt of this output indicates that the entire synchronization sequence has been received via the corresponding propagation path, and that the next symbol representation associated with that propagation path is a representation of the first data symbol.

As the symbol representations arrive at matched filters 120, they also arrive simultaneously at RAM 126, where each arriving symbol representation is stored in successively addressed storage locations. The addresses for successive storage are indicated by an address pointer under the control of controller 124. Whenever RAM 126 indicates to controller 124 that another symbol representation has arrived and has been stored, controller 124 increments the address pointer by one symbol so that the next symbol representation to arrive will be stored at the next address in the sequence. The storage in RAM 126 is cyclical: after a symbol representation has been stored in the last address of RAM 126, the next symbol representation to arrive is stored in the first address of RAM 126, overwriting whatever had been stored there previously.

Controller 124 also notes when signals are received from threshold detectors 122. This enables controller 124 to infer the relative times of arrival, via the various propagation paths, of the incoming symbol representations (after the last symbol of a synchronization sequence arrives, one data symbol of the corresponding propagation path arrives every T_s subsequently), and to make a table of which addresses in RAM 126 correspond to the same symbol. As soon as all the symbol representations of the first several symbols are received, starting with the first symbol of data sequence 118, controller 124 retrieves those representations from RAM 126 and sends them to amplitude estimator 98, delay unit 100 and phase estimator 102, with the symbol 16

representation (h_λ,$_λ) corresponding to the propagation path of highest amplitude (index /=λ) being sent first. (The propagation path of highest amplitude is identified based on amplitude averaging over several symbol durations T_s. This operation can be performed easily during the acquisition phase.) Subsequently, as soon as all the symbol representations of a particular symbol are received, controller 124 retrieves those representations from RAM 126 and sends them to amplitude estimator 98, delay unit 100 and phase estimator 102, with symbol representation (/?_λ,φ_λ) being sent first for each symbol.

Figure 10 illustrates the storage in RAM 126 of representations of symbols arriving via three different propagation paths a, b and c, and the subsequent output of those symbol representations. Propagation path a is the fastest path. Propagation path c is the slowest path. The data symbols are numbered sequentially starting from 1. The last symbols of the synchronization sequence are labeled with question marks (?). The storage locations in RAM 126 are labeled sequentially with their addresses, starting from address 0.

In the particular case illustrated in Figure 10, the first representation of the last symbol of the synchronization sequence is stored in the first storage location, whose address is 0. The address pointer then is incremented so that the next arriving symbol representation is stored in the storage location with address 1. Note that the last two representations of the last synchronization symbol arrive after the first representation of the first data symbol arrives, but before the second representation of the first data symbol arrives, so that symbol representations b? and c? are shown stored at addresses 4 and 5, between symbol representation al at address 3 and symbol representation bl at address 6. Also in the particular case illustrated in Figure 10, propagation path a, in addition to being the fastest propagation path, also is the propagation path of highest amplitude. The output sequence has the representations of each symbol grouped together, with the symbol of largest amplitude first: al, bl and cl for the first symbol, a2, b2 and c2 for the second symbol, and so on. As noted above, the storage in RAM 126 is cyclical. For example, if RAM

126 has 128 storage locations, numbered 0 through 127, after symbol representation a42 has been stored at address 127, symbol representation c41 is stored at address 0, symbol representation b42 is stored at address 1 , and so on. In general, the number of storage locations in RAM 126 must be at least as great as the product of the number of paths L and the number of subsequences in the pseudonoise sequence. If it is desired to have access to the symbol representations of the synchronization sequence along with the representations of the first data symbol, then, to ensure that the synchronization sequence is not overwritten, the minimum number of storage locations in RAM 126 is L(M+K), where M is the number of subsequences in the pseudonoise sequence and K is the number of symbols in the synchronization sequence. Optionally, controller 124 checks for error conditions, for example the receipt of a trigger from a threshold detector 122 later, relative to the first trigger, than the length of the pseudorandom noise sequence. The data symbol representations of the associated propagation path then are ignored, and are not forwarded to RAKE combiner 104. In short, the embodiment of symbol synchronizer 96 that is illustrated in

Figure 9 provides an elegant solution to the two problems of:

(a) identifying the representations of the first data symbol, and

(b) grouping together the representations of the same data symbol for processing by RAKE combiner 104. The reason for ordering the symbol representations with the symbol representation of highest amplitude first is to enable RAKE combiner 104 to combine the symbols coherently with reference to the phase of the symbol representation associated with the propagation path of highest amplitude. The coherent combination of the symbol representations that is produced by RAKE combiner 104 is y = h_λg_λ exp jφ_λ + ∑h,g, exp[y^'(^, - Δθ_u )] l≠Λ where h_t and φ are respectively the amplitude and phase of the received symbol representation for propagation path /, originally supplied by correlator bank 86 and properly time- aligned by delay unit 100. As above, g, is an estimate of the true amplitude associated with propagation path / and received from amplitude estimator 98. →θ_lλ is an estimate of the difference between the true phase of propagation path / and the true phase of propagation path λ, received from phase estimator 102: ^A u = ^θι - θ_λ

Δθ_lλ is calculated by averaging successive values of φ_rφ_λ. Phase wrap is accounted for by making the calculation in the proper phase interval. The advantage of this combining algorithm over the prior art algorithms is that although the phase of each representation is affected by the modulation, the phases of different representations of the same symbol are affected identically. As a result the difference between two phases of two representations of the same symbol changes much more slowly, from symbol to symbol, than do the phases themselves. Therefore, the phase difference estimator of the present invention is more accurate than the absolute phase estimator of the prior art. This is particularly important, because, with the total signal energy spread among several propagation paths, the signal-to-noise ratio of any one path is low.

In other words, RAKE combiner 104 of the present invention rotates all the phases to the reference phase of the symbol representation corresponding to the propagation path of highest amplitude.

If the symbols of data sequence 118 have been produced by differential modulation (e.g., DBPSK or DQPSK), then symbol detector 48 or 106 recovers the bits of the original message merely by multiplying each symbol received from RAKE combiner 46 or 104 by the complex conjugate of the immediately succeeding symbol. For DBPSK, the value of the bit is 1 if the product lies in the first or fourth quadrant of the complex plane, and -1 if the value of the product lies in the second or third quadrant of the complex plane. If the symbols of data sequence 118 have been produced by non-differential modulation (e.g., BPSK or QPSK), then symbol detector 48 or 106 should include a mechanism for estimating the reference phase of coherent combination y. This reference phase is subtracted from the phase ofy, and the value of the difference determines the sign of the bit. The estimation of the reference phase can be done in several ways, for example, the VIVA algorithm cited above. Note that these algorithms are more robust for this purpose than for the purpose of estimating the absolute phases of the individual symbol representations, because coherent combination y has a higher signal-to-noise ratio than the separate representations. While the invention has been described with respect to a limited number of embodiments, it will be appreciated that many variations, modifications and other applications of the invention may be made.

Claims

WHAT IS CLAIMED IS:

1. A receiver for a signal that includes a plurality of symbols modulated with a pseudonoise sequence and that travels to the receiver along a plurality of paths in a channel, comprising:

(a) a matched filter for correlating the signal with the pseudonoise sequence to produce a correlation function;

(b) a delay estimator for estimating a delay for each of the paths from said correlation function; and

(c) a plurality of correlators for correlating the signal with the pseudonoise sequence in accordance with said delays.

2. The receiver of claim 1 , further comprising:

(d) a mechanism for combining outputs of said correlators to recover the symbols.

3. The receiver of claim 2, further comprising:

(e) a channel estimator for estimating a plurality of parameters of the channel; and wherein said mechanism is operative to combine said outputs in accordance with said parameters.

4. The receiver of claim 3, wherein said plurality of parameters includes, for each path, an amplitude and a phase.

5. A receiver for a signal that includes a plurality of symbols modulated with a pseudonoise sequence and that travels to the receiver along a plurality of paths in a channel, comprising:

(a) a plurality of correlators; and

(b) a controller for configuring said plurality of controllers to alternately: (i) correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function, and (ii) for each of the paths, correlate the signal with at least a portion of the pseudonoise sequence in accordance with a delay associated with the path.

6. The receiver of claim 5, further comprising:

(c) a delay estimator for estimating said delays from said correlation function.

7. The receiver of claim 5, further comprising:

(d) a mechanism for combining outputs of said correlators, when said correlators are configured to correlate the signal with the pseudonoise sequence in accordance with said delays, to recover the symbols.

8. The receiver of claim 7, further comprising:

(e) a channel estimator for estimating a plurality of parameters of the channel; and wherein said mechanism is operative to combine said outputs in accordance with said plurality of parameters.

9. The receiver of claim 8, wherein said plurality of parameters includes, for each path, an amplitude and a phase.

10. In a communications system in which a message including a plurality of symbols is modulated with a pseudonoise sequence and the resulting signal is transmitted along a certain number of propagation paths, a method of receiving the signal and recovering the symbols, comprising the steps of:

(a) providing a receiver including: (i) a matched filter and

(ii) a plurality of correlators;

(b) acquiring the signal, using said matched filter; and

(c) tracking the signal, using at least one of said plurality of correlators.

11. The method of claim 10, further comprising the step of:

(d) estimating the number of propagation paths; said tracking being effected using at least a like number of said correlators.

12. The method of claim 11, wherein said acquiring includes the step of correlating the signal with at least a portion of the pseudonoise sequence to produce a correlation function including a certain number of peaks, said estimated number of propagation paths being said number of peaks.

13. The method of claim 11 , further comprising the step of:

(e) estimating, for each propagation path, a delay; said tracking being effected by correlating the signal with at least a portion of the pseudonoise sequence in accordance with each said delay, using at least one corresponding said correlator for each said delay.

14. In a communications system in which a message including a plurality of symbols is modulated with a pseudonoise sequence and the resulting signal is transmitted along a certain number of propagation paths, a method of receiving the signal and recovering the symbols, comprising the steps of:

(a) providing a receiver including a plurality of correlators;

(b) configuring said plurality of correlators to correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function;

(c) acquiring the signal, using said correlation function;

(d) configuring at least one of said correlators to track the signal; and

(e) tracking the signal, using said at least one correlator.

15. The method of claim 14, further comprising the step of:

(f) estimating the number of propagation paths; said tracking being effected using at least a like number of said correlators.

16. The method of claim 15, wherein said correlation function includes a certain number of peaks, and wherein said estimated number of propagation paths is said number of peaks.

17. The method of claim 15, further comprising the step of: (g) estimating, for each propagation path, a delay; said tracking being effected by correlating the signal with at least a portion of the pseudonoise sequence in accordance with each said delay, using at least one corresponding said correlator for each said delay.

18. A method of transmitting and receiving a message including a plurality of symbols, all of the symbols having a common symbol duration, comprising the steps of:

(a) providing a pseudonoise sequence;

(b) modulating the symbols with said pseudonoise sequence to produce a signal;

(c) providing a receiver including: (i) a matched filter and

(ii) a plurality of correlators;

(d) acquiring the signal using said matched filter; and

(e) tracking the signal using at least one of said correlators.

19. The method of claim 18, wherein said pseudonoise sequence has a duration longer than said symbol duration.

20. The method of claim 19, wherein said pseudonoise sequence includes a plurality of subsequences, each of said subsequences having a duration equal to said symbol duration.

21. The method of claim 20, wherein said acquiring is effected by steps including correlating the signal with a first said subsequences.

22. The method of claim 18, further comprising the step of: (f) estimating a set of propagation parameters of said signal.

23. The method of claim 22, wherein said set of propagation parameters includes a number of propagation paths, said tracking being effected using at least a like number of said correlators.

24. The method of claim 22, wherein said set of propagation parameters includes a maximum delay, and wherein said pseudonoise sequence has a duration greater than said maximum delay.

25. A method of transmitting and receiving a message including a plurality of symbols, all of the symbols having a common symbol duration, comprising the steps of:

(a) providing a pseudonoise sequence;

(b) providing a receiver including a plurality of correlators;

(c) configuring said plurality of correlators to correlate the signal with at least a portion of the pseudonoise sequence to produce a correlation function;

(d) acquiring the signal, using said correlation function;

(e) configuring at least one of said correlators to track the signal; and

(f) tracking the signal, using said at least one correlator.

26. The method of claim 25, wherein said pseudonoise sequence has a duration longer than said symbol duration.

27. The method of claim 26, wherein said pseudonoise sequence includes a plurality of subsequences, each of said subseqences having a duration equal to said symbol duration.

28. The method of claim 25, further comprising the step of: (g) estimating a set of propagation parameters of said signal.

29. The method of claim 28, wherein said set of propagation parameters includes a number of propagation paths, said tracking being effected using at least a like number of said correlators.

30. The method of claim 28, wherein said set of propagation parameters includes a maximum delay, and wherein said pseudonoise sequence has a duration greater than said maximum delay.

31. In a communications system wherein a message including a plurality of symbols is transmitted to a receiver along each of a plurality of propagation paths, the transmission along each propagation path being effected with a certain delay, the message including a preamble and a data sequence, the preamble terminating in a synchronization sequence, the receiver producing, for each symbol of the data sequence and for each symbol of the synchronization sequence, a plurality of representations of that symbol, each representation corresponding to a unique one of the propagation paths and being produced at a time corresponding to the delay of the corresponding propagation path: a method for identifying and grouping the representations of each symbol of the data sequence for further processing, comprising the steps of:

(a) sequentially storing the representations, as the representations are produced, in a plurality of sequentially addressed locations;

(b) identifying a first representation of a first symbol of the data sequence;

(c) for each subsequent representation of said first symbol, identifying a corresponding address increment; and

(d) for each symbol of the data sequence, retrieving the representations of that symbol together in accordance with said address increments.

32. The method of claim 31, wherein said identifying of said first representation of said first symbol of the data sequence is effected by steps including collectively identifying first representations of the symbols of the synchronization sequence.

33. The method of claim 32, wherein said collective identifying of said first representations of the symbols of the synchronization sequence is effected using a matched filter.

34. In a communications system wherein, for each of a plurality of symbols, a plurality of representations is obtained, each representation having an amplitude and a phase, a method of combining the representations of each symbol, comprising the steps of: for each symbol:

(a) identifying, from among the representations, a reference representation;

(b) for each other representation:

(i) estimating a true difference between the phase of said each other representation and the phase of said reference representation, and

(ii) adjusting the phase of said each other representation in accordance with said estimated true difference, thereby producing a phase-adjusted representation; and

(c) summing said phase-adjusted representations with said reference representation, thereby producing a coherent combination of the representations.

35. The method of claim 34, wherein said reference representation is a representation associated with a propagation path of highest amplitude.

36. The method of claim 34, wherein said estimating of said true difference between the phase of said each other representation of said each symbol and the phase of said reference representation of said each symbol is effected by steps including:

(A) subtracting the phase of said reference representation of said each symbol from the phase of said each other representation of said each symbol, thereby providing a phase difference for said each symbol;

(B) for at least one other symbol: 27

(I) identifying, from among the representations of said at least one other symbol, a reference representation of said at least one other symbol, and

(II) subtracting the phase of said reference representation of said at least one other symbol from the phase of a representation of said at least one other symbol corresponding to said each other representation of said each symbol, thereby providing a phase difference for each of said at least one other symbol; and (C) averaging said phase difference of said each symbol with said at least one phase difference of said at least one other symbol.

37. The method of claim 34, wherein said adjusting of the phase of said each other representation is effected by steps including subtracting said estimate of said true difference from the phase of said each other representation.

38. The method of claim 34, further comprising the steps of: for each symbol: prior to said summing:

(e) estimating a true amplitude of each representation of said each symbol; and

(f) adjusting the amplitude of said each representation in accordance with said true amplitude.

39. The method of claim 38, wherein said estimating of said true amplitude is effected by steps including averaging the amplitude of said each representation with the amplitude of a corresponding representation of at least one other symbol.

40. The method of claim 38, wherein said adjusting of the amplitude of said each representation is effected by steps including multiplying the amplitude of said each representation by said estimated true amplitude.

41. The method of claim 34, further comprising the step of, for each symbol:

(d) recovering, from said coherent combination, a bit corresponding to said each symbol.

42. The method of claim 41, wherein said recovering is effected by steps including multiplying said coherent combination by a complex conjugate of an immediately succeeding coherent combination.

43. The method of claim 41, wherein said coherent combination has a phase, and wherein said recovering is effected by steps including:

(i) estimating a reference phase for said coherent combination; and (ii) subtracting said reference phase from said phase of said coherent combination.