MXPA06002173A - Method and apparatus for improving channel estimate in presence of short spreading codes - Google Patents

Abstract

A method and apparatus for estimating a communication channel impulse response h(t)is disclosed. The method comprises the steps of generating com(t)=co(t + mNTcc) for m=0,1,&Lgr;, Mby correlating a received signal r(t) with a spreading sequence Siof length N (208), wherein the received signal r(t) comprises a chip sequence cjapplied to a communication channel characterizable by an impulse response h(t)(204), and wherein the chip sequence ci is generated from a data sequence di spread by the spreading sequence Si(202);generating an estimated communication channel impulse response hM (t)as a combination of com(t)and dm form=0,1,&Lgr;,M (210);and filtering the first estimated communication channel impulse response hM(t)to generate the estimated communication channel impulse response h(t)with a filter f selected at least in part according to the spreading sequence Si (212).

Description

"METHOD AND APPARATUS TO IMPROVE CHANNEL CALCULATION IN THE PRESENCE OF SHORT DISPERSION CODES" FIELD OF THE INVENTION The present invention relates to systems and methods for communicating information, and in particular to a system and method for calculating the impulse response of a communications channel using short synchronization codes.

BACKGROUND OF THE INVENTION In packet-based communication systems, dispersion codes are used for packet detection and synchronization purposes. Correlation techniques are used to identify and synchronize their timing. In many cases, the dispersion code sequence may be in the order of 1000 chips or more. Since the receiver must be correlated in all possible delays, this process can result in unacceptable delays. To improve this problem, a short spreading code with good aperiodic auto-correlation can be used for packet detection and synchronization purposes. An example is the IEEE 802.11 Wireless Local Area Network (WLAN) system, which uses a Barker code of 11 long symbols as a scatter sequence for the preamble and header of a packet. The short length of the scattering sequence makes it easy for receivers to quickly detect the presence of a packet in the communications channel and to synchronize to its timing. In the case of a linear channel, for receiver design purposes, it is often desirable to calculate the impulse response of the communications channel. In the context of WLAN, a multi-path linear channel is often used, and such communication channels require equalization for effective reception. Given a calculation of the impulse response of the communication channel, we can directly calculate the equalizer coefficients by matrix calculations, contrary to conventional adaptive algorithms. This is described in "Digital Communications," by John G. Proakis, 4th edition, August 15, 2000, the reference of which is incorporated herein by reference. This allows to calculate the equalizer coefficients in a digital signal processor (DSP) instead of a more expensive and less adaptable dedicated hardware that implements adaptation algorithms.

Unfortunately, because the dispersion code used is short (for example, of the order of 11 symbols) a direct correlation using the scatter code will produce a distorted calculation. What is needed is a simple, computationally efficient technique that can be used to calculate calculations < pulse response of substantially undistorted communications channel, even when the received signal had been microprocessed with a short spreading code. The present invention satisfies that need.

BRIEF DESCRIPTION OF THE INVENTION In order to address the requirements described above, the present invention describes a method and apparatus for calculating a pulse response of the communications channel h (t). The method comprises the steps to generate com (t) = co (t + mNTc) for m ~ 0.1, A, M by correlating a received signal r (t) with a dispersion sequence Si of long N, where the signal received r (t) comprises a sequence of chips Cj applied to a communications channel characterized by a pulse response h (t), and in which the chip sequence Cj is generated from a sequence of data scattered by the dispersion sequence YES; generating a calculated impulse response of the communications channel rjM (t) as a combination of de com (t) and dm for m-0,1,?, M; and filtering the first calculated impulse response of the communication channel f? M (t) to generate the calculated impulse response of the communications channel h (t) with a filter f selected at least in part according to the dispersion sequence Si . The apparatus comprises a correlator to generate com (t) = co (t + mNTc) for m = 0, 1,?, M when correlating a received signal r (t) with a dispersion sequence Si of long N, where the received signal r (t) comprises a sequence of chips Cj applied to a communications channel characterized by a response of impulse h (t), and where the sequence of chips Cj is generated from a sequence of data d dispersed by the sequence of dispersion Si! a calculation to generate a calculated impulse response of the communications channel fiM (t) as a combination of com (t) and dm for m = 0.1,?, M; and a filter f selected at least in part according to the dispersion sequence S ±, the filter for filtering the first calculated impulse response of the communication channel ÍtM (t) to generate the calculated impulse response of the communication channel h ( t). The above allows the impulse response h (t) of the communication channel to be calculated accurately, even with short chip codes. or intuitively, in the case of a channel impulse response limited in time, the present invention provides a calculation that can be made perfect at the limit of the high signal-to-noise ratio (SNR).

BRIEF DESCRIPTION OF THE DRAWINGS Referring now to the drawings in which the reference numbers represent corresponding parts along the same: Figure 1 is a diagram of a transceiver system; Figure 2 is a block diagram illustrating the process steps that can be used to implement the present invention; Figure 3 is a diagram of a transceiver system using a filter f to improve the calculated impulse response of the communication channel; Figure 4 is a diagram showing the response of the filter; Figure 5 is a flow chart describing exemplary processing steps that can be used to improve reconstruction of the impulse response value of the communications channel using supercodes imposed on the portion of the data stream; Figure 6 is a diagram of a transceiver system using supercode to transmit sequences; Figure 7 is a diagram showing a correlator output using the Barker code of 11 long symbols; Figure 8 is a diagram showing a correlator output using Walsh codes as an input supercode; Figure 9 is a diagram showing a correlator output after post-processing with a filter f as described in Figure 2 as Figure 3; Figure 10 is a diagram showing a more detailed view of the main lobe peak, showing the calculation of the impulse response of the communication channel in the current impulse response of the communication channel; and Figure 11 is a diagram showing a mode of a processor that can be used to carry out the present invention.

DETAILED DESCRIPTION OF THE INVENTION In the following description, reference is made to the accompanying drawings that form a part thereof, and which are shown, by way of illustration, of various embodiments of the present invention. It is understood that other embodiments and structural changes may be used without being insulated from the scope of the present invention.

System Model Figure 1 is a diagram of a transceiver system 100. Using a signal scatterer 103, a sequence of random data symbols di 102, comprising a series of data packets 128 (each of which may include a preamble 124 used by the receiver for identification purposes, as well as a data payload 126) is dispersed by a sequence Si 104 long N:. { Sn, O = n = N-1} and that has a period of chips. The sequence Si 104 is known by the receiver 112 a priori. The dispersed chip sequence Cj 105 is, therefore: Cj = clmn = di • Sn, 0 < n = N- 1 Eq. (1) This chip sequence Cj 106 is transmitted by a linear transmission channel 108 having a combined channel impulse response h (t). The transmitted signal is received by a receiver 112. The received waveform r (t) 114 is: °° r (í) =? C, • h. { t - jTc) + n (t) Eq. (2) where n (t) 121 is an additive noise component. This formulation does not explicitly impose a causality requirement on h (t) 108. If explicit causality is desired, this can be done by setting h (t) = 0, t < 0. For purposes of simplicity, all data and code sequences in the following description are assumed to be real, however the channel impulse response h (t) 108 and the additive noise component n (t) 121 could be complex in their baseband representations. Complex sequences could easily be accommodated if needed, but are not common for synchronization purposes. The receiver 112 receives the transmitted signal, and correlates the received signal r (t) 114 with the known spreading sequence Si 104 to identify the data as intended to be received by the receiver 112. Once the received signal r (t ) 114 is received, the preamble can be examined to determine the direction of the data and if additional processing is necessary. Such systems also use the received signal to calculate the input response of the communications channel 108. This information is used to improve the detection and subsequent reception of signals from the transmitter 110. In circumstances where the scattering sequence S¿ 104 is relatively short, the data packet 128 must be detected quickly, and there is less data available to calculate the response of communication channel 108.

Conventional Detection and Synchronization For detection and synchronization purposes, the dispersion code search is conventionally performed by correlating the received signal r (t) 114 with the dispersion sequence. This is done by the correlator 116. Although this correlation is typically performed after sampling in the time domain, for the simplicity of the notations, it does not execute the discretization of the time domain. The output of correlator 116 co (t) 118 is determined by: ? M? R (t + iTc). If Ec. (4; i = Q (t) Eq. (5; Eq. (6) = jD (í) «? (í-ffJ + íi (í) Ec. (7) where D (t) is the correlation between the chip sequence and the dispersion sequence and we will refer to it as the chip correlation. For simplicity of notations, we have introduced a group delay (negative). { 1TC) when calculating output 118 of the correlator. The output of the correlator 116 is determined by the convolution of the chip correlation D (l) with the impulse response of the sampled communications channel h (t-lTc) plus a noise component ñ (t). After the additional examination: D (/) =? C / + /. 5 / Eq. (8) =? dm • S "+ l • S, +? dm + l • Sn + i_N • Sl Ec. (9) / = 0 i = N-n = dm * A. { n) + dm ^ * A. { N-n), Eq. (10) 1 = mN + n, 0 < n < N Eq. (11) where A (n) is an aperiodic self-correlation of two parts of the dispersion sequence defined as: A (n) is a property of the code sequence that is known by the correlator 116 a prior !.

For detection and synchronization purposes, the scattering sequence S¿ 104 is designed to have minimum values of A (k) when k? 0. However, for small values (eg, of the order of 10) of N (short scatter codes), even the smallest lateral lobe magnitude is not negligible compared to in-phase self-correlation. The Barker sequences, when they exist, provide the best aperiodic self-correlation. ' For a Barker sequence of 11 chips, S¿ = 1, -1,1,1, -1,1,1,1, -1, -1, -1, the self-correlation becomes A (i) = 11.0, -1.0, -1.0, -1.0, -1.0, -1 for 0 = i < ll. Note that even for Barker codes, because the scattering sequence S¿ 104 is limited in length, the auto-correlation A (t) includes significant side lobes. The output 118 of the correlator 116 can be rewritten as: i) .h (t- { jN + i) Tc) + ñ (t) Eq. (13) .A (N-l)). H (t- (jN + i) Tc) + ñ (t) Eq. (14) t-. { jN + i) Tc) + A. { N-l) * h (t - ((jN + i) Tc)) + ñ (t) Eq. (15) it-JNTe-iTe) + Ec. (16) =? dj »fÍ (t-jNTc) + ñ (t) Ec. (17) where the following is defined as the convolution of the aperiodic auto-correlation of the dispersion sequence A (i) and the channel impulse response sampled h (t-iTc) as explained below: ? M í? (T) =? A (i) .h (t-iTc) Eq. (18); =? G + I This is a calculation of the impulse response h (t) of the combined communication channel 108 to the output of the code correlator 116. The above equations can be written more succinctly using a convolution notation. Define a convolution of two infinite sequences Ai and Bi as C = A®BoC (i) =? A (j) * B (i-j), \ / i Eq. (19) By defining an OtT operator that converts any O sequence into a function in the time domain that uses the Dirac delta function: Bt (t) =? B (j) * d (t-iT) Eq. (20) we can also define the convolution of a function with a sequence that uses a normal convolution of two functions: C (t) = A (t) ® Bt C (t) =? A (t- T) B (j) Ec. (21) Using the above notations and in addition, adopting the following definitions: u (iN) = di (data) Ec. (22A) u (iN + n) = * 0,0 < n < N Eq. (22B) Sn, 0 = n < N S (n) =. { "Ec. (22C) 0, other (a limited-time microprocessing sequence) The preceding equations (1), (2), (3), (6), (12), (18), (16), (17) can be rewritten as: c = u ® S Eq. (1 ') r = h ® ctTc + nc Eq. (2') co = r ® StT_ Ec. (3 ') = __ ® ctT ® Strc_ + ñ, where S_ (n) = S (-n) Ec. (6 ') A = S®S_ Ec. (12 ') £ = h ® Arrc Ec. (18 ') co = h®ut c + ñ Eq. (16 ') = ^ ®drwrc + ñ Eq. (17 ') Determine a Communication Channel Impulse Response Calculation For the simplicity of the notations, in the remaining description, we assume that the data symbols are binary. However, the results can generally be applied to non-binary data. Because the correlator 116 has access to the same sequence of codes S¿ 104 used to generate the dispersion chip sequence c-106 prior to transmission, the correlator 106 can correlate the received signal r (t) 114 with the sequence code Si 104. However, duplication with short code sequences Si 104 may occur, because time delays may cause correlator 116 to correlate different portions of adjacent code sequences. Conventionally, these duplication effects are reduced by integrating or adding multiple periods of codes (eg, M), as described above. As described in Equations (13) - (17), based on the output 118 of the correlator 116 we can form a calculation of the channel impulse response over a code period _TC: =) +? d0 • dj • ñ (t - jNTc) + d0 • ñ (t)? o where do is a value of the data at time t = 0. This is a rough approximation of h (t), corrupted by duplicate copies of it, (t) spaced in multiples of NTC of the desired copy. These terms of duplication and additive noise can be reduced by adding in M code periods: i M-l (t) = -? d, (t + mNTc) Eq. (24) M m = 0 M-l -kt) + ~ ?? dm * dj.f ?. { t + (m-j) NTc) + ñM (t) Ec. (25) M m-Oj? M i M-l: ^ + ~ ?? m • d m .ñ. { t-INTc) + ñM (t) Ec. (261 The foregoing indicates that by means of the output 122 of the calculator 120, by eliminating the data modulation by correlating with the data sequence, we obtain a calculation ii of the channel impulse response plus the terms defined by the auto-correlation of the data sequence, which disappears when infinite terms are added. If DM (1) is defined as: i M-l DM. { l) = -? dm »dl + m Ec. (27) so . NTc f? M = h®DM + ñM Eq. (28) where ftM ss a calculation of the impulse response of the communications channel h (t). When the data sequence di 102 is random, white and independent of the additive noise n (t) 121, and in the limit of M? 8: D ~ (l) = * d? 0, ñ '-, = 0 Ees. (29) Therefore, at the limit of the infinite sum (as M approaches infinity), we obtain a calculation that is equal to the true channel impulse response h (t) in convolution with the aperiodic self-correlation of the dispersion sequence Yes 104. As demonstrated by the above, we can not obtain the true channel impulse response h (t) with simple integration. The best we have is covered by the self-correlation of the dispersion sequence S¿ 104. In cases where the dispersion sequence Si 104 is long, the self-correlation approaches a delta function, and the lateral lobes disappear. However, when the dispersion sequence S ± 104 is short, the lateral lobes of the self-correlation are not insignificant and • will cause significant distortion for the calculation of the true channel impulse response h (t).

Improved Channel Calculations for Short Scattering Sequences As demonstrated below, the present invention improves the communication channel impulse response calculation by filtering the first calculated impulse response of the communication channel f? M to generate the calculated response of pulse of the communications channel h (t) with a filter f selected at least in part according to the dispersion sequence Si. In particular, when the time interval of the communications channel 108 is limited, a forced deconvolution to zero can be used to improve the calculation. Figure 2 is a block diagram illustrating the process steps that can be used to implement the present invention. Figure 3 is a diagram of a transceiver system 300 using the filter f described above to filter the first calculated impulse response of the communications channel where f ΔM (t) to generate an improved calculation suitable for short spreading sequences Yes 104. Referring to Figure 2 and Figure 3, blocks 202 to 208 describe the steps that are used to generate com (t) 118. A sequence of scatter chips Cj 106 is generated from a sequence of data symbols di 102 and a dispersion sequence Si 104 long N, as shown in block 202. The chip sequence Cj 106 is transmitted via a communication channel 108 as shown in block 204, and is received as shown in block 206. The communication channel includes the transmitter 110 and the receiver 112. The received signal r (t) 114 is then correlated with the scattering sequence Si 104 by the correlator 116 to generate com (t) as shown in block 208. In block 210, a computed impulse response of the communications channel where ñM (t) by the calculator 120 as a combination of com (t) and dm for m = 0.1, K, M. This can be done, for example, using the relationship described in Eq. (24) discussed above. Finally, in block 212, the first response of the computed communications channel where τM (t) is filtered with a filter f selected at least in part according to the dispersion sequence Si 104. In one embodiment, the filter is a filter f 302 finite impulse response (FIR) designed with the following restrictions: Af = A® f Ec. (29) Af (0) = 1, Af (n) = 0.0 <; | n | = L Eq. (30) where A < ? > f is the convolution of the self-correlation of the dispersion sequence S¿ 104 and the filter, and Af is the self-correlation of the scattering sequence S¿ 104 after filtering. Figure 4 is a diagram showing the response of the filter f 302 described in Equations (29) and (30).

When the calculation of the impulse response of the communication channel is filtered with this filter, we obtain: hf = f? ®f tTc t Tc t Tc = h®A ®f Ec. (31) tTc = h®Af Using this technique, the effects of the lateral lobes (duplicate versions of the self-correlation of the dispersion sequence S 104) are eliminated between L and -L. The side lobes are not completely eliminated (since the filter passes components greater than L and smaller than -i) but the result is close to the origin. { n = 0) is of primary interest, and the effect of the lateral lobes can be significantly reduced in this region. If the time interval (duration of the impulse response) of the communications channel is less than LTC, that is: 3tj < t2, t2-t1 < LTc, / t < tl Jt > t2:? (f) «0 Eq. (32) (ie, there is a time t2 greater than tx that defines a time interval t2-t? less than LTC, and for all time outside the interval t? -ti, h (t) is close to zero), After, the filtered calculation hf (or, in the previous notation, hf (t)) is composed of an exact copy of h (h (t)), plus some duplicate versions of it in non-overlapping positions. So in this case h has resolution from hf. Such a filter with long 2L + 1 can be designed with the criteria of simple forced to zero: L? A (n-i) »f f) = Af n), -L = n = L Ec. (33) i = -L where f (i) is the impulse response of the filter f 302 in such a way that Af. { n) is a convolution of A (n) and f (i), Af. { n) = l for n = 0 and Af [n) = 0 for 0 < | n | = ¿, y? Mn A (n) = A (-n) =? S¡ • Sl + n, 0 = n = N and where N is a length of the = or sequence of chips Si 104. L it can be selected such that the LTC product (the chip period Tc is known) is approximately equal to the time interval (e.g., the approximate duration of the impulse response) of channel 108. Note that the value A (ni) is well defined ... is a property of the dispersion sequence If 104, which is known a prior i. As usual, the matrix structure of the linear equations is Toeplitz. Due to the design requirement of the dispersion sequence Si, the matrix must be well conditioned. The filter coefficients can be calculated off-line given the dispersion sequence and the desired window width L. Although the above has been described with respect to non-recursive filters, other filters, such as recursive filters, can also be used. For example, a recursive filter can provide a perfect filtering of the side lobes, but the result may not be the conditioned matrix, consequently the solution may be more difficult to determine. In fact, any filter of length 2L + 1 can be defined.

Supercoded Transmission Sequences It has been shown that given f? and with filtering, it is possible to recover the true channel impulse response for a limited channel of time. However, in the above description, it was obtained. by integration for multiple periods of dispersion sequence. The number of periods you need to integrate can be large especially if 2X = N, since we rely on self-correlation data to suppress duplicate copies of%. In one embodiment of the present invention, supercodes, such as Walsh-type supercodes, are used to drastically reduce the amount of integration required. This technique is especially useful in systems that have a sufficient signal-to-noise ratio (SNR). Consider a couple of Walsh codes of length 2 wO =. { + 1, + 1 } and Wi =. { + 1, -1} . These codes can be used to form a sequence of data: Any segment of 2 long symbols in this sequence can be described as either w0 or -WQ, except for a single w? in the middle. If this sequence is now correlated with w_., The resulting correlation will be characterized by a single peak in the center and zeroes in any other place (except near the limits). The negatives of the two codes can be taken (for example, w0 =. { -1, -1} and w? =. { -1, -1-1} and / or their roles can be exchanged (eg, = { +!, +.}., and w0 = { + 1, -1.} with the same result, the three additional patterns are listed below thus obtained and their correlator patterns: ..., -, -, -, -, +, +, +, + ... -, + - / - t + ~ t - r + ~ r + ~ r - r + ~ r - ••• + ~ r + ~ ..., +, -, +, -, - / +, -, + ... Since the following results are equivalent for all previous patterns when additive noise it is not correlated in the sampling points, we limit our description to the first data sequence (that is, ..., +, +, +, -, -, -, ...) In this case, f? t (t) = - (dQ • co (t) + d »co (t + NTC)) Ec. (36) 2 = f? (T) +? (Dl -dM) »f? (T-INTc) + n2 (t) Ec. (37) /? O = ft (t) +? { dt -dM f? (t-INTc) + n2. { f) Ec. (38) / < /,? / = / 2 If the condition can be satisfied that -l N > (2N + __) I 12N > (2N + L), fi can be reconstructed free from duplication interference, and by deconvolution (filter technique mentioned above), h can also be reconstructed. From the foregoing, it can be determined that a small supercode imposed on a portion of the data stream can provide a free duplication calculation of the impulse response of the communication channel when the channel response is time limited. The only source of distortion from this calculation comes from the additive noise, which can be suppressed by the number of times of the scattering gain by a factor of 2 (to answer for the supercode). When noise is low, such an approach is preferable over large integrations. For moderate values of L, such code sequences can easily be incorporated into a longer preamble to pack data, probably with multiple copies, without adversely affecting the spectral properties of the transmission. Furthermore, when the signal-to-noise ratio (SNR) is low, the traditional integration delineated in the first half of this section can still be performed in such a preamble in order to obtain a high processing gain against additive noise. Figure 5 is a flowchart describing the exemplary processing steps that can be used to improve the reconstruction of the impulse response value of the communications channel using the supercode imposed on a portion of the data stream. Figure 6 is a diagram of a transceiver system 600 that uses super-coded transmission sequences to generate an improved pulse response calculation of the communications channel suitable for short scatter sequences Si 104. In block 502, a sequence is generated data stream 102. The data stream 102 includes one or more data packets 128, each data pack having a preamble 124 that includes a restricted portion Cdi 602. The preamble 124 may be, for example, in the form of pseudo code random.

The restricted portion Cdi 602 is associated with at least two codes, w0 and w_ .. The codes w0 and W? are selected in such a way that the code correlation () of the restricted portion Cdi 602 and at least one of the codes wQ and w ?, is characterized by a maximum value at k = 0, and is worth less than the maximum value in k? 0. Ideally, the ACode correlation (k) of the restricted portion Cdi 602 is a pulse, with ACode (k) equal to one in k = 0, and equal to the other values for k. However, because such correlation characteristics are typically achievable, the codes wo and w_. can be selected to approximate this ideal. For example, the codes WQ and wi can be selected such that the correlation ACode (k) of the restricted portion Cdi 602 and at least one of the codes wo and wí r is such that ACód ± go (k) = 1 in k = 0 and AC (say (k) * 0 for substantially all k? 0. 0, the codes w0 and w_ can be selected such that the Andigo correlation (k) of the restricted portion Cdi 602 and at least one of the codes WQ and w_., is such that ACode (k) = 0 for 0 <| k | = J, where J is selected to minimize the correlation of the restricted portion Cdi with one of the codes WQ, W? for substantially all k 0 0. In one embodiment, the restricted portion Cdi 602 comprises the pair of Walsh codes of length two in the first sequence described above.They encompass other modalities in which the codes are of another length (different from length two) , or they are different codes to a Walsh code. In block 504, a sequence of chips Cj 106 is generated. The chipset Cj 106 is generated by applying a dispersion sequence Si 104 of long N_ and having a chip period Tc to the data sequence di 102. This sequence of scatter chips Cj 106 is transmitted by a transmission channel 108. linear that has a combined channel impulse response h (t). The transmitted signal is received by a receiver 112. In block 506, the receiver 112 receives the transmitted signal, and correlates the received signal r (t) 114 with the known dispersion sequence Si 104 in order to identify the data as it is posed to be received by the receiver 112. This is done by generating com ( t) = co. { t + mNTc) for? a = 0.1, A, M, using techniques analogous to those described above. In block 508, a calculated impulse response of the communications channel f? M (t) is generated as a combination of the correlation com (t) and the data sequence dm for m = 0.1, A, M. In one mode, the codes w0 and w_ are two Walsh codes with long two symbols, and? it is calculated as, with M = 2. In this case, 1 M M - ll -? dm. co. { t + mNTc) M (equals f2 (t) = - (d? • co (t) + dl »co (t + NTC)) Therefore, where data has been restricted with a symbol such as a Walsh supercode, a Improved calculation of the impulse response of the communication channel can be obtained by taking two consecutive values of the correlation of the received data and the dispersion sequence and multiplying each result by the data sequence In the example of Walsh codes w0 = { -1, -1.}. And WQ = { -1, -1.}. Applied to the sequence ..., +, +, +, -, -, -..., and Wi applied to the receiver, the result is that one of the values of co (t) is multiplied by one, and the other is multiplied by minus 1. Consequently, the output will produce essentially no response until the transition between the two codes occurs. of Walsh, at which time a clean copy, without duplication, of the impulse response of the communications channel will be produced. A supercode of length 2 has been described for a Improved suppression of duplication. When the SNR is low and a longer integration period is desirable, it would seem more attractive to generalize the code in longer lengths. Counterintuitively, this is not possible. This result is shown below, presenting a definition of such codes and showing that there is no code with a length greater than 2 for the binary data sequences. An infinite sequence A forms a pair of impulsive correlation with a finite sequence B of length L if A satisfies the following equations: A (i) = B (i), V0 < < L-l? A (i + n) »B (i) = 0, VH? O i = 0 In counterpart, it can be seen that for binary sequences, such a pair does not exist for L > 2 . Assuming that such sequences exist, it is apparent that L must be even. Considering two such cases (i = 4k and L = 4k + 2) In the first case, L - 4k, consider the first restriction: L- \? A (i-l) .B (i) = 0, Eq. (39a)? = 0 L-l A { - \) • 7J (0) +? B (i-1) • 73 (7) = 0 Eq. (39b); = 1 Since there are 4k addends in the equation that carries the values from. { + 1, -1} half of them or 2k terms must be positive, and the other half negative. The product of all the addends must be consequently 1.

A (-l). B (L-1) = l Ees. (40) Similar arguments can be used to show that: A (i) = B (L + i), - £ < i < 0 Eq. (41) But this implies that: L-l L-l? ? A (.i - L)) • 5B ((0i) == A ((- IL)) • 5B ((00)) ++ ?? _ 4 (/ - Z) • B (i); = 0 == 1? L ~ \ A (-Z) .5 (0) +? 5 («5 (0 Eq. (421 = ^ (- Z) »5 (0) + X-l > 0 which contradicts the assumption that the cross-correlation is zero anywhere except at the origin. Therefore, in opposition, we have shown that for binary sequences, such a pair does not exist for L > 2 . A similar argument can be applied for the second case, L = 4k + 2, except that the product of all the addends in each equation must be -1, since we must now have 2k + l negative terms. This leads to: A (i) = (-l) and B (L + i), - £ < í < 0 Eq. (43) When k > 0 L-l L-l? A i-2) • 5 (0 = -4 (-2) • 5 (0) + A { - \) • 5 (1) +? A i-2) • 5 (0 i = Q = 2 L-l (Z-2) «5 (?) ~ 5 (Z-l)« 5 (l) +? 5 (.- 2) «5 (0 = 0 L-l = 5 (Z-2) .5 (0) +5 (Z-l) «5 (l) +? (- lY5 (.- 2) .5 (0 = 0 Ees. (44) Adding the two equations together we have: 2A 5 (X-2) = 5 (0) + S? 5 (2 / -!) • 5 (20 = 0 Eq. (45) / = 1 However, this result is clearly impossible since there is an odd number of terms to the left. Consequently, in opposition, it is shown that it is impossible to satisfy the restrictions when L > 2 for binary sequences.

Effects of Noise The above has shown that the distortions due to this dispersion sequence design can be eliminated from the calculation of the impulse response of the communications channel. Now pay attention to the remaining distortion caused by the additive noise, n (t) 121. Assuming that the noise source is white and stationary and that it is filtered by a receiver filter for bandwidth coupling, its measurement of Distortion can be defined as explained below: t Tc ñu = ñM ®f Ec. (47: i L M-l = - ?? dm. (t + mNTc -ITc) .f (l) Eq. (48; L M-l -Í7 ??? dm * n. { t + mNTc + iTc -lTc) .S (i) .f (l) M i = L m = 0 i 1 M-l F? S *. • n. { t + mNTc + jTc) .Rfs. { j) M m = 0 j where Rfs (J) =? FJ s (i + j) Eq. (49); = - L The joint expectation of Eq. (46) can take care of n (t), whose self-correlation can be determined by the filter of front reception and supposedly known). (50) _ 1 M ~ l Rß U) = -? Dm »Rp0 + mN) Ec. (5i: When the noise n (t) is white, we have: Rm (k c) =, vk? or EXAMPLES Figure 7 to Figure 10 are diagrams illustrating the performance improvements achieved by the application of the present invention. These illustrated the examples and places them, so that a Barker code of length 11 is used as the scattering sequence S¿ 104. Figures 7-10, magnitudes normalized as a function of the chip timing. No adjustment is made for the group delays introduced by the correlation, filtering and windowing, consequently the time coordinates must be treated in a relative sense. Figures 7-10 also do not include the effects of additive noise. Figure 7 is a diagram showing a correlator output 116 that uses a Barker code of length 11 and conventional pulse response calculation techniques of the communication channel. The output of the correlator 116 shows a main lobe peak 702, and multiple spurious peaks 704. These spurious peaks 704 (which are at 11 chips, or NTC seconds, apart due to the Barker code of length 11) are due to transmission repeated of the short code Si 104, which "duplicate" one after the other. If the length of the periodic scattering sequence Si 104 were longer, there would be fewer spurious peaks 704, and the peaks 704 would not overlap the main lobe peak 702 as much as shown in Figure 7. Figure 8 is a diagram showing a correlator output 116 that uses the codes of Walsh in conjunction with the supercode technique described in Figure 5. To generate this graphic representation, the input data was restricted with the Walsh codes of two symbols of long w0 and w_., And the output was processed by adding two outputs successive of correlator 116 as shown in Eq. (36). For the 11 chips on either side of the main lobe peak 702, there is zero correlation, and many of the spurious correlator peaks 704 that were apparent in Figure 7 are no longer evident. However, note that since only six bits of the data sequence are restricted ... +, +, +, -, -, -..., some duplicate versions of the main lobe peak 704 (labeled 802) are present (at 33 chips of the main lobe peak 702). However, since these duplicate versions 802 are well separated from the main lobe peak 702, an accurate calculation of the impulse response of the communications channel can be obtained. Note that a similar result can be achieved without restricting the input sequence with the supercode, but this would require integration over a large number of symbols (for example, M in Eq. (26) would be large). Note also that the main lobe peak 704 still includes minor peaks due to the calculator 120 producing fi, which is a dirty version of h. These undesirable components 804, caused by the self-correlation of the dispersion sequence 104, can not be eliminated by restricting the data sequence. In contrast, these undesirable components 804 can be removed by filtering as described with respect to Figure 9 shown below. Figure 9 is a diagram showing an output of the correlator 116 shown in Figure 8 after post-processing with a filter f as described in Figures 2 and 3. Note that the side lobes 802 shown in Figure 8 are shown in FIG. have rejected the main lobe peak 702, and some of the undesirable components 804 of the main lobe peak 702 have been filtered. Also note that the indexing of data (the chips shown as the time axis) of Figure 9 has changed in relation to the data indexing in Figure 8. As described above, this difference is an artifact of the software used to graphically represent Figure 7-Figure 11 and is not associated with the applicant's invention. Figure 10 is a diagram showing a more detailed view of the main lobe peak 702, showing the computation of the impulse response of the communication channel (indicated by the asterisks) and the current impulse response of the communication channel. Note that the calculated impulse response of the communications channel proceeds the current response very closely.

Hardware Environment Figure 11 is a diagram illustrating a processor system 1102 by way of example that could be used in the implementation of selected elements of the present invention (including, for example, portions of the transmitter 110, the receiver 112, the correlator 116). , the calculator 120, or the filter 302).

The processor system 1102 comprises a processor 1104 and a memory 1106, such as a random access memory (RAM). Generally speaking, the processor system 102 operates under the control of an operating system 1108 stored in the memory 1106. Under the control of the operating system 1108, the processor system 1102 accepts input data and commands and provides output data. Typically, instructions for executing such operations are also incorporated into an application program 1110, which is also stored in memory 1106. The processor system 1102 may be incorporated into a microprocessor, a desktop computer, or any similar processing device. The instructions implementing the operating system 1108, the application program 1110, and the compiler 1112 can be tangibly incorporated into a computer readable medium, for example, the data storage device 1124, which could include one or more storage devices. of fixed or removable data, such as a zip handler, a floppy disk drive, hard disk drive, CD-ROM driver, tape driver, etc. In addition, operating system 1108 and application program 1110 are comprised of instructions which, when read and executed by computer 1102, cause computer 1102 to perform the steps necessary to implement and / or use the present invention. Application program 1110 and / or operating instructions can also be tangibly incorporated into memory 1106 and / or communication devices 1130, thereby making an application program product or article of manufacture according to the invention. As such, the terms "article of manufacture", "program storage device" and "computer program product" as used herein are intended to encompass a computer program accessible from any device or computer-readable medium. . Those skilled in the art will recognize that many modifications can be made to this configuration without being insulated from the scope of the present invention. For example, those skilled in the art will recognize that any combination of the above components, or any number of different, peripheral components, and other devices, may be used with the present invention. For example, an application-specific integrated circuit (ASIC) or a Field Programmable Gate Array (FPGA) can be used to implement selected functions, including the correlator 116, and the Filtering functions can be executed by a general-purpose processor, as described above.

Conclusion This concludes the description of the preferred embodiments of the present invention. The above description of the preferred embodiment of the invention has been presented for purposes of illustration and description. It is not intended to be exhaustive to limit the invention to the precise form described. Many modifications and variations are possible by virtue of the previous teaching. It is intended that the scope of the invention be limited not by this detailed description, but by the claims appended thereto. The above specification, examples and data provide a complete description of the manufacturer and use the composition of the invention. Since many embodiments of the invention can be made without being isolated from the spirit and scope of the invention, the invention resides in the appended claims below.

Claims (57)

1. NOVELTY OF THE INVENTION Having described the invention as antecedent, the content of the following claims is claimed as property CLAIMS 1. A method for calculating a pulse response of the communications channel h (t), characterized in that it comprises the steps for: generating com (t) = co (t + mNTc) for I? = 0.1, A, M by correlating a received signal r (t) with a scattering sequence S¿ of long N, where the received signal r (t) comprises a sequence of chips cj- applied to a communication channel characterized by a pulse response h ( t), and where the chip sequence c3 is generated from a data sequence di dispersed by the scattering sequence 5_. and where Tc is the chip period of the chip sequence Cj, generating a calculated impulse response of the communication channel hM () as a combination of com (t) and dm for m = 0.1,?, M; and filtering the first calculated impulse response of the communication channel ñM (t) to generate the calculated impulse response of the communication channel h (t) with a filter f selected at least in part according to the scattering sequence Sa, 2 The method according to claim 1, characterized in that the filter f is further selected at least in part according to a self-correlation A (n) of the scattering sequence Sj .. 3. The method according to claim 2, characterized in that the filter f is further selected at least in part according to the duration of the impulse response of the communications channel h (t). 4. The method according to claim 2, characterized in that the filter f is additionally selected at least in part according to a criterion L from forced to zero? (A (n - i) • f (i)) = Af (), -L = n = L, where:? = - L f (i) is the impulse response of filter f such that Af (n ) is a convolution of A (n) and f (i); Af (n) = 1 for n = 0 and Af (n) = 0 for 0 < \ n \ = L, and N - n Af (n) = A (-n) =? S, • Sl + ", 0 = n = N, and N is a; = 0 over the sequence of chips S. The method according to claim 4, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communication channel h (t) is less than LT. The method according to claim 4, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communications channel h (t) is approximately equal to LTC. 7. The method according to claim 1, characterized in that N is less than 20. The method according to claim 1, characterized in that M = 0. 9. The method according to claim 1, characterized in that the data sequence di includes a restricted portion Cdi associated with at least two Wo codes. WÍ, where an Andigo (k) correlation of the restricted portion of the restricted portion Cdi with one of the codes Wo, w? is characterized by a maximum value in k = 0 less than the maximum values in k? 0 The method according to claim 9, characterized in that the step to generate a calculated impulse response of the communications channel HM (t) as a combination of com (t) and dm for m = 0,1,, M comprises i MM --ll the step to calculate f? M (t) as - dm • co (t + NTc) 11. The method according to claim 10, characterized in that = 2. The method according to claim 9, characterized in that the data sequence di includes a preamble having a pseudo-random code that includes the restricted portion of the data sequence di. The method according to claim 9, characterized in that Andigo (k) = 1 at k = 0 and Ac? Aigo (k) = 0 for substantially all k? 0. The method according to claim 9, characterized in that Andigo (k) = 0 for 0 < | k | J, where J is selected to minimize the correlation of the restricted portion Cdi with one of the codes w0, Í for substantially all k? 0. The method according to claim 14, characterized in that 2J is a length of the restricted portion Cdi. 16. The method according to claim 1, characterized because Aco-code (k) = 1 in k = 0 and Code (k) = 0 for substantially all k? . 17. The method according to claim 1, characterized in that each of the two codes w0, W? It comprises two symbols. 18. The method according to claim 1, characterized in that each of the two codes 0, w? It comprises no more than two symbols. The method according to claim 1, characterized in that the codes 0, comprise Walsh codes. 20. An apparatus for calculating a pulse response of the communications channel h (t), characterized in that it comprises: means for generating com (t) = co (t + mNTc) for m = 0.1, A, M when correlating a received signal r (t) with a dispersion sequence Si of long N, where the received signal r (t) comprises a chip sequence cj applied to a communication channel characterized by a pulse response h (t), and where the chip sequence Cj is generated from a data stream di dispersed by the scattering sequence Si and where Tc is the chip period of the comedian chip sequence to generate a calculated impulse response of the communication channel f? M. { f) as a combination of com (t) and dm for m = 0.1,?, M; and filter means f, selected at least in part according to the dispersion sequence If r the filter means for filtering the first calculated impulse response of the communication channel M (t) in order to generate the calculated impulse response of the communications channel h (t) with 21. The apparatus according to claim 20, characterized in that the filter means f is additionally selected at least in part according to a self-correlation A (n) of the dispersion sequence Si. The apparatus according to claim 21, characterized in that the filter means f is additionally selected at least in part according to the duration of the impulse response of the communication channel h (t). 23. The apparatus according to claim 21, characterized in that the filter means f is additionally selected at least in part according to a criterion L from forced to zero. { (n-?) »f (i)) = Af. { n), -L = n = L, where:? = - L f (i) is the impulse response of the filter medium f such that Af (n) is a convolution of A (n) and fd); Af (n) = 1 for n = 0 and Af (n) = 0 for 0 < | n | = ¿; and N - \ - n A (n) = A (-n) =? S, »S¡ + n, O = n = N, and N has a 7 = 0 length of the Si chip sequence. 24. The apparatus according to claim 23, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communication channel h (t) is less than LTC. 25. The apparatus according to claim 23, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communication channel h (t) is approximately equal to LTC. 26. The apparatus according to claim 20, characterized in that N is less than 20. 27. The apparatus according to claim 20, characterized in that M = 0. The apparatus according to claim 20, characterized in that the data sequence di includes a restricted portion Cdi associated with at least two codes w0, w ?, where an ACodxgo correlation (k) of the restricted portion Cdi with one of the codes w0 , w? is characterized by a maximum value in k = 0 less than the maximum values in k? 0. 29. The apparatus according to claim 28, characterized in that the means for generating a calculated impulse response of the communication channel f? M. { t) as a combination of com (t) and dm for m = 0, l,?, M comprises means to calculate / i (/) as X. 'j • Co (t + mNT) • M -1 m 30. The apparatus according to claim 29, characterized in that M = 2. The apparatus according to claim 28, characterized in that the data sequence di includes a preamble having a pseudo-random code that includes the restricted portion of the data sequence di. 32. The apparatus according to claim 28, characterized in that Acó ± go (k) = 1 at J = 0 and Andigo (k) = 0 substantially for all k? O. 33. The apparatus according to claim 28, characterized in that ~ Acó-digo (k) = 0 for 0 < | k | = J, where J is selected to minimize the correlation of the restricted portion Cdi with one of the codes W0, WÍ for substantially all k? 0. 34. The apparatus according to claim 33, characterized in that 2J is a length of the restricted portion Cdi. 35. The apparatus according to claim 20, characterized in that Andigo (k) = 1 in k-0 and Code (k) = 0 for substantially all k? O 36. The apparatus according to claim 20, characterized in that each of the two codes w0, w? It comprises two symbols. 37. The apparatus according to claim 20, characterized in that each of the two codes w0, w? It comprises no more than two symbols. 38. The apparatus according to claim 20, characterized in that the codes wo, w_. they include Walsh codes. 39. An apparatus for calculating a pulse response of the communications channel h (t), characterized in that it comprises: a correlator that generates com (t) = co (t + mNTc) for m = 0.1, A, M when correlating a received signal r (t) with a scattering sequence Si of long N, where the received signal r (t) comprises a chip sequence Cj applied to a communication channel characterized by a pulse response h (t), and where the chip sequence Cj is generated from a sequence of scattered data. by the scattering sequence Si and where Tc is the chip period of the chip sequence Cj, - a calculator for generating a calculated impulse response of the communication channel f? M. { f) as a combination of com (t) = and dm for J? = 0.1, A,; and a filter f selected at least in part according to the dispersion sequence Si, the filter for filtering the first calculated impulse response of the communication channel f? M (t) in order to generate the calculated impulse response of the channel of communications h (t). 40. The apparatus according to claim 39, characterized in that the filter f is further selected at least in part according to a self-correlation A (n) of the scattering sequence Si. 41. The apparatus according to claim 40, characterized in that the filter f is further selected at least in part according to the duration of the impulse response of the communication channel h (t). 42. The apparatus according to claim 40, characterized in that the filter f is additionally selected at least in part according to a criterion L from forced to zero. { (ni) »f (f)) = Af (n), -L = n = L where: t = -L f (i) is the impulse response of filter f such that Af (n) is a convolution of A (n) and f (i); Af (n) = 1 for n = 0 and Af (n) = 0 for 0 < | n | < i; and N-l-p A (n) = A (-n) =? S,; »Yes +", O = n = N, and N is a long / = 0 of the chip sequence S¿. 43. The apparatus according to claim 42, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communication channel h (t) is less than LTC. 44. The apparatus according to claim 42, characterized in that: the parameter L is selected in such a way that a duration of the impulse response of the communications channel h (t) is approximately equal to LTC. 45. The apparatus according to claim 39, characterized in that N is less than 20. 46. The apparatus according to claim 39, characterized in that M = 0. 47. The apparatus according to claim 39, characterized in that the data sequence di includes a restricted portion Cdi associated with at least two codes w0, w ?, where a correlation ACode (k) of the restricted portion Cdi with one of the two codes w0, w? is characterized by a maximum value in J = 0 less than the maximum values in k? 0. The apparatus according to claim 47, characterized in that the calculator for generating a calculated impulse response of the communications channel hM (t) as a combination of com (t) and dm for m = 0.1,?, M i Ml comprises means to calculate fu (t) as -? dm * co (t + mNTc)49. The apparatus according to claim 48, characterized in that M = 2. 50. The apparatus according to claim 47, characterized in that the data sequence di includes a preamble having a pseudo-random code that includes the restricted portion of the data sequence di. 51. The apparatus according to claim 47, characterized because ACode (k) = 1 in k = 0 and Acode (k) = 0 for substantially all k? O. 52. The apparatus according to claim 47, characterized in that ACode (k) = 0 for 0 < | k | = J, where J is selected to minimize the correlation of the restricted portion Cdi with one of the codes w0, W? for substantially all k? O 53. The apparatus according to claim 52, characterized in that 2J is a length of the restricted portion Cdi. 54. The apparatus according to claim 39, characterized in that Acodíg0 (k) = 1 in k = 0 and ACode (k) = 0 for substantially all k? O 55. The apparatus according to claim 39, characterized in that each of the two codes w0, W? It comprises two symbols. 56. The apparatus according to claim 39, characterized in that each of the two codes w0, i comprises no more than two symbols. 57. The apparatus according to claim 39, characterized in that the two codes w0, W? they include Walsh codes.