MXPA99000213A - Adaptive communication system and method using unequal weighting of interference and noise - Google Patents

Abstract

A communication system and method, particularly adapted for use in adaptive antenna arrays, are presented. A plurality of signals containing a desired signal, interference, and noise are received through an array of antennas connected to a base receiving station. The signal arriving at each antenna is weighted according to computations which adjust those weights as a function of received interference and noise power in an unequal manner, based on the estimation accuracy of the interference and noise, which in turn depends on factors such as interference to noise power ratios and fading rate.

Description

ADAPTABLE SYSTEM AND METHOD OF COHUNICATION USED »AN UNEQUAL WEIGHTING OF INTERFERENCE AND NOISE FIELD OF THE INVENTION The invention relates to the field of communication technology, and particularly to receiving stations having arrays of multiple antennas, the signals received through which they are selectively combined to produce an output signal.

BACKGROUND OF THE INVENTION It is known in the art of communications that receiving stations equipped with antenna arrays, instead of a single antenna, can improve receiver operation. Antenna arrays can reduce the multipath fading of the desired signal and suppress interference signals. Such arrangements can consequently increase the range and capacity of wireless systems. This is true for example for wireless cellular phones and other mobile systems. In mobile systems, a variety of factors cause signal corruption. These include the interference of other mobile users in or near a given cell. Another source of signal degradation is multipath fading, in which the REF. 29202 amplitude and phase received from a source varies with time. The fading speed can reach up to 200 Hz for a mobile user traveling at 60 mph (95.6 kph) at PCS frequencies of approximately 1.9 GHz. In such environments, the problem is to cleanly extract the signal from the user being tracked from the collection of noise, interference and portions of the desired signal received added in the array antennas. It is also known in the art that after receiving that signal collection, one way of extracting the desired signal is to take the signal above each array antenna, multiply this by a weighting factor, then add the weighted signals or adjusted from each of the antennas to produce the output signal. The desired signal to interference plus noise (SINR) ratio is increased as much as possible in the summed output. This helps to achieve an accurate reception and, in digital signals, reduces the percentage of error in bits (BER). Since antenna arrays offer recognized advantages, including a greater total power of the received signal, a key aspect is the optimal calculation of the weights used in the array. Different methods for the generation of weights in the technique have been presented. If the channels of the wanted and interference signals are known, the weight generation technique maximizes the SINR, which also minimizes the average square error (MMSE) between the output signal and the desired output signal, is the well-known equation by Weiner-Hopf, w - IR »] - 1 rxd, (1) where R ^ a denotes the cross-relation of the vector of the received signal -e with the desired signal, given by r? d - E [x d (2) where d is the desired signal, and RM is the correlation matrix of the received signal, which in turn is defined as R ^ - E [x ^ \ (3) where the supra-index * denotes the complex conjugate and T denotes the transposition. It is important to note that the received signal consists of the desired signal, noise and interference, and thus (3) can also be expressed as Rxx - PJU U /, (4) where Pd is the power of the desired signal u * is the vector of the channel for the desired signal, s2 is the power of the noise, Pj and Uj are the power and the channel vector of the jth interference, and L is the number of interferers. A variation of this weight equation is given by where Ri + n - E [(x-r xdd) * x-r ^ d) 7] (4) These weights differ only by a scalar factor of (1), and therefore the output SINR is the same. However, these previous techniques represent only an idealized case, given that the channel is assumed to be stationary and it is assumed that perfect knowledge of the channel characteristics (R ** and rxd) arrive at the calculation of the weights. When the characteristics of the channel are not stationary and unknown, the standard method in the prior art is to estimate R1 + n and r? A of the received signals, for example, using a rectangular window of K samples, the weights at time k are given by, w (k) = k + - (k) r ",. { k) (7) where where that test symbol denotes the estimates of R? + n, rx, and d. This method uses the estimated values (those are the estimates of maximum probability with K samples in a stationary environment) for R? + N, -. Xd * and d in the Weinwe-Hopf equation, as if the channels were actually stationary and those values were exactly known. This technique is known in the literature as Direct Matrix Investment (DMI) or matrix investment sampled. This DMI technique however produces weights which deviate from the ideal ones (MMSE) for at least two reasons. The first source of degradation is the propagation of the error. To estimate the desired signal in (9), the DMI algorithm can use an instruction sequence, but if the channel varies between the instruction sequences, then you must also use estimates of the desired signal data. The standard known method is to use the detected data of the output signal. However, errors in the detection cause errors in the weights, increasing the maximum BER, and possibly leading to a propagation of the error through the receiver system which may last until the next instruction sequence. The second cause of degradation is that the characteristics of the channel may vary with the duration of the window of K symbols. In this way, the size of the window should remain as small as possible to preserve the accuracy of the estimates. However, in the DMI it is the computational case in which the estimates are noisy due to the finite length of the window. In the DMI, the degradation of the SINR due to noise depends on the ratio of K to the number of weights (M), which is also the number of antennas. For example, when _ / A ^ 2, that loss is approximately 3 dB and the loss increases with M. Due to this loss, the increase in the number of antennas (with a fixed K) can lead to a decrease in the operation when the weights are estimated using the DMI, which is contrary to the effect with the ideal weights where the operation always increases with M. Since this error of estimation of the weight can cause a great degradation in the output SINR in comparison With the case of ideal weight, methods have been proposed to reduce this degradation in the prior art for specific interference situations. If it is assumed that there is only noise, and not interference, then (5) is reduced to the weights of the combination of the maximum ratio (MRC) given by w = rx < / (10) and the weights can be calculated by "(*) = '", (*) 01) which ignores any interference. If it is assumed that there is a very strong interferer that dominates the effect of noise, then one's own analysis can be used effectively, by the weights given by / w (k) --- (/ - «, (k)«, (*)) / • "(*) (12) which completely cancels the interference regardless of the noise level. Note that this algorithm operates with weights that are orthogonal to the subspace of the interference. Although (11) and (12) give a much better performance than the DMI when there is no interference and when there is a dominant interference, respectively, they can get much worse than the DMI when assuming untrue conditions- that is, that (11) degrades operation more when the level of interference is increased, and (12) degrades operation more when the level of the single interferer decreases in relation to noise and other interference. In general, both techniques worsened much more than DMI when noise and interference have similar powers, which is often the case in wireless communication systems. All prior weight generation techniques (DMI, MRC and proprietary analysis) for antenna arrays are known in the art, and each places the extraction of the desired signal to a greater or lesser degree, depending on the conditions. An additional discussion of them can be found in the literature, including in the articles: J.H. Winters, "Signal Acquisition and Tracking with Adaptive Arrays in the Digital Mobile Radio System IS54 with Flat Fading", IEEE Transactions on Vehicular Technology, Vol. 42, No. 4 (November 1993); J.H. Winters et. al, "The Impact of Antenna Diversity on the Capacity of Wireless Communication Systems", IEEE Transactions on Communications, April 1994, and US Patent Nos. 4,639,914 entitled "Wireless PBX / LAN System With Optimum Combining", and 5,481,570 entitled "Block Radio and Adaptive Arrays for Wireless Systems ", both of JH Winters, each article or patent including one of the inventors of the present application are incorporated herein by reference. None of the above methods, however, has achieved that they would be considered adequate results under a wide range of communication conditions. Known systems suffer from disadvantages, including proper operation only under specific known conditions, poor operation when those conditions are not maintained, and poor operation in the most common conditions of an interference and noise mixture. Therefore, there is a strong need for a system that works well under all conditions of interference and noise, even without any prior knowledge of the environment, ie a weight calculation technique that adapts itself to the interference environment .

BRIEF DESCRIPTION OF THE INVENTION The invention that overcomes these and other problems in the art is an aspect of an adaptive communication system and method for employing controllable weights for channels associated with multi-antenna receiver arrays. The communication system and method of the invention employ a weighting scheme in which the weights are generated using estimated values for interference and / or noise, which are adjusted unevenly, depending on the accuracy of their estimation. In the best performance, including but not limited to when the results of interference and noise are almost equal. In another aspect the invention provides a communication system and method which vary the weights according to the powers of noise and interference, that is, the prevailing communication conditions. In another aspect the invention provides a communication system and method which reduce the dependence of the estimated interference or noise, when the errors in those estimates increase. In another aspect the invention provides a communication system and method which can be adapted to improve upon DMI, self-analysis and other previously known arrangement techniques.

In another aspect the invention provides a communication system and method which improves the performance of wireless cellular phones and other mobile communication systems. The invention relates more particularly to the typical application of a weighting factor (0) to the noise component and different weighting factors (ßt to | 3M) to the components of the interference, although it will be appreciated by those skilled in the art. technique that different mathematical forms of unequal weighting could be used.

BRIEF DESCRIPTION OF THE DRAWINGS The invention will be described with reference to the following drawings: Figure 1 illustrates an antenna array receiver system, in schematic form. Figure 2 illustrates the components of the receiving station, in the invention. Figure 3 illustrates a schematic example of the invention with an array of four antennas in operation. Figure 4 illustrates graphs indicating the operation of a communication system according to the invention, and according to conventional systems.

DETAILED DESCRIPTION OF THE PREFERRED MODALITIES As illustrated in Figure 1, the communication system and method of the invention relate to an antenna array consisting illustratively of a plurality of generally coupled antenna elements, 20? up to 20M >; Each element of the array antenna receives RF signals and passes the received signals through a group of corresponding channels 30? up to 30M. As illustrated in Figure 2, channels 30? up to 30M in turn are connected to processor units 40, which typically contain a CPU 50, an electronic memory 60, a fixed store 70, circuits and signal processing and related programs, connected with the appropriate collective conductor and other circuits, such as it will be appreciated by those skilled in the art. The processing unit 40 is constructed to determine and apply the weights necessary to carry out the system and method of the invention, as described herein. As noted above, in the JMl and other receiving systems of the prior art, errors in the weights generated can be introduced because those methods use estimates, instead of the ideal values (which are generally not known in the receiver) , for the correlation matrix Rxx or Ri- "and the correlation vector of the desired signal rxd, in the weight equations that assume ideal values. The inventors have made the following observations, which helped to motivate their method of inventive communication. First, if the BER is sufficiently low, as is typically the case in a wireless communication system that operates properly, the main cause of the degradation of the SINR is the estimation error of Rxx or Ri + n, instead of rx. In addition, and as can be shown, the estimation error of Rxx grows with M, while the estimate value of rxd depends on? F. Third, Rxx is composed of the desired signal, noise, and interference, as shown in (4), while j + "is composed of noise, and interference. Specifically, R. + n can be expressed in terms of its eigenvectors, (13) • I where? i and ßi are the iési or eigenvalue and eigenvector of R? + n respectively, where «i results from the orthogonalization of the vectors of the interferer Uj. Further, (13) can be expressed in terms of noise and interference as Fourth, the ideal weight equation (5) was based on an equal effect of noise and interference on the SINR. However, the estimation error of the eigenvectors of the interference may differ from that of the noise. For example, the noise level can be known very accurately (as is the thermal noise of the receiver, which typically does not vary), while the eigenvalues and eigenvectors of the interferent can have a substantial error of estimation. Also, the errors in the weights due to the estimated interference may have a greater effect on the degradation of the SINR than the errors due to the noise, and the interference, which is a signal which is not as random as an analog of the signal desired, can affect the BER differently from ordinary noise. In addition, the estimation error of the eigenvectors of the interferer (e ±) generally depends on the strength of the interferer, with the strong interferers estimated with greater accuracy than the weak interferers.

In this way, equal weighting of noise and interference (as in (7)) with the weight estimate it can not give the lowest BER weights achieved, even though the unequal weighting could increase the BER in the absence of the weight estimation error.

Therefore, the inventors in their communication system and method propose a weight generation method that uses unequal weighting of noise and interference. The best unequal weighting will depend on the noise and interference estimation error, which in turn depends on the strength of the interferers in relation to the noise, as well as other parameters such as the fade rate,,, and M. This error of Estimation does not depend on the strength of the desired signal, though, as seen in (14). Using unequal weighting, in the communication system and method of the invention 10 the generation of the weight is given by w (* > < A, s' l +? ßiúi -s' ftYr '-, (*) (15) where e and X are the same eigenvector and eigenvalue of Ri + n (A :) It is advantageous for the ß¿. { i > 0) are a function of the X., that is, the level of interference, for example, ß * c (? -s Y, where c eS a constant already increases with the estimation error, that is, that increases with the fade rate and M / K. Various modalities of the system and method of communication of the invention are of interest Note that (15) requires the calculation of all eigenvectors and eigenvalues of "Si + n (Jc) , which can be computationally intensive Therefore, the inventors propose to reduce the complex versions of (15) that require the calculation of less or none of the eigenvectors and eigenvalues.The simplest implementation is to weight the noise unevenly with total interference In this case, (15) can be expressed as which does not require the calculation of the eigenvectors or eigenvalues of "+" (.-) From the foregoing, extending (16) to the algorithm of the MMSE weight, * > - ((* J *) + Ar1"* &U?) According to the results obtained by the inventors, when the noise interference ratio (INR) is very large, ß »0 gives the lowest BER. On the contrary, when the INR is very small, β-1 gives the lowest BER. As illustrated schematically in Figure 3, the communication system and method of the invention can be applied to a receiving station having 4 antennas (Af = 4), in which there is an interference signal source 1 Si, together with the source of the desired signal S0. If it is known that the INR in such communication facility is small, then ß can be set at 0.75, so that the noise is weighted three times stronger than the interference. The inventors have also determined that in the intermediate case, with INR between the polar ends, there is a range of ßs, 0 <; ß < l, which gives the BER lower than the DMI or MRC methods, with the greatest improvement over those known techniques when the powers of interference and noise are comparable. The inventors have verified that the optimum ß (for the lowest BER) is independent of the desired signal-to-noise ratio (SNR), and depends on INR and fading speed (e.g., vehicle speed). Therefore, in a set of illustrative conditions, consider the optimum β for the worst case of fade rate of 184 Hz. Thus, the optimal β depends mainly on the INR, which decreases with the increase in TNR. This is because the INR determines the estimation error in the weights to suppress the interference. The inventors have determined that the BER is relatively insensitive to small variations around the optimal β. Given this insensitivity, an exact estimate for the optimal β or INR function (or I / N, in dB) under these conditions is 1 HN = - \ Q 20- - fN -10 < // tf = S 30 '"SKIIN = * (18) 23- A 8 < /.Vs23 30 23 < // ?.

INR can be known in advance, for example in noise or limited interference environments, but typically INR is not known and varies with fading. For the Therefore, the inventors also propose to use as an estimate value of the INR. Note that to calculate tr [Ri + n], only the diagonal elements R1 + n are necessary, and thus | *. (19) where Xi and rxdj are the ith elements of x and rxcj, respectively. Thus, in the system and method of the invention an algorithm (without parameters to be fixed in the field) can be used to estimate INR, for all environments. Alternatively, if INR is known in advance, for example if it is known that INR varies over a small interval or if an INR design is desired for the worst case, then a constant ß can be used, as determined for this INR. Figure 4 shows the BER versus the SIR for the previous method with SNR = 10 dB with a fade of 184 Hz. The results for DMI, MRC, and ideal weights are also shown. The invention works for EMU and MRC for all SIRs. In another modality, only a higher eigenvalue of? I and the corresponding eigenvector mt of Ri + n are considered, which corresponds to the stronger interferent. This corresponds to a typical wireless system where there is a dominant interferent. Based on (15), the system of the invention in this case can be implemented by constructing ceno weights: w (k) = [. { 1-ßo-) R? + N (?) + > gbs2I + A (?? - s2) é *? ^ ^ r ^ ik) (20) As above, the inventors have determined that their method provides a substantial improvement over the proprietary analysis (as well as DMIA and MRC) with a single interferer, particularly when the interfering power is not dominant with respect to noise. Also, as previously the ßi can be obtained from tr [R? + R ..}. , but in this case the ßi can also be obtained from the larger eigenvalue of Ra + ", (corresponding to the eigenvector" i). The above description of the system and method of communication of the invention is illustrative, and variations in its construction and implementation will occur to those skilled in the art. For example, although illustrative embodiments are described in terms of an antenna array whose antenna elements are spatially diverse, the invention could be adapted to systems in which the diversity of signals received is temporal in nature, i.e., using a temporary equalizer. instead of the space equalizer discussed above. Similarly, the technique can be used with spatial and temporal equalization together. The invention, therefore, is intended only to be limited to the following claims.

It is noted that With respect to this date, the best method known to the applicant to carry out the aforementioned invention, is that which is clear from the present description of the invention. Having described the invention as above, property is claimed as contained in the following:

Claims (39)

1. A method for processing a plurality of received signals, characterized in that it comprises the steps of: generating a plurality of weights based on a ratio of the power of a desired signal to an equally weighted sum of the noise power and the interference power; and weighting and combining the plurality of signals received using the plurality of weights generated to amplify the desired signal and suppress the interference signals.

2. The method according to claim 1, characterized in that the unequally weighted sum is based on the accuracy of the estimate of an interference.

3. The method according to claim 1, characterized in that the unequally weighted sum is based on an interference to noise ratio.

4. The method according to claim 3, characterized in that the interference to noise ratio is an estimate.

5. The method according to claim 3, characterized in that the interference to noise ratio is a known value.

6. The method according to claim 3, characterized in that the relation is an estimate.

The method according to claim 1, characterized in that the power of the interference comprises a plurality of unequally weighted interference powers.

The method according to claim 7, characterized in that the unequally weighted interference powers are estimates.

9. The method according to claim 7, characterized in that the unequally weighted interference powers are based on unequal interference powers.

10. The method according to claim 9, characterized in that the individual interference powers are estimates.

The method according to claim 2, characterized in that the unequally weighted sum is also based on a fade rate.

12. The method of complying with claim 1, characterized in that the plurality of received signals are received from a plurality of spatially diverse receiving elements.

13. The method according to the claim 1, characterized in that the plurality of received signals are received from a plurality of temporally diverse receiving elements.

14. An apparatus for processing a plurality of received signals, characterized in that it comprises: a weight generating unit, for generating a plurality of weights based on a ratio of a power of the desired signal to an unequally weighted sum of the power of the noise and the power of interference; and a combination unit, for weighting and combining the plurality of signals received using the plurality of weights generated to amplify the desired signal and suppress the interference signals.

The apparatus according to claim 14, characterized in that the unequally weighted sum is based on the accuracy of an interference estimate.

The apparatus according to claim 14, characterized in that the unequally weighted sum of the noise power and the interference power is based on an interference to noise ratio.

17. The apparatus according to claim 16, characterized in that the interference to noise ratio is an estimate.

18. The apparatus according to claim 16, characterized in that the interference to noise ratio is a known value.

19. The apparatus according to claim 16, characterized in that the relation is an estimate.

20. The apparatus according to claim 14, characterized in that the power of the interference comprises a plurality of unequally weighted interference powers.

21. The apparatus according to claim 20, characterized in that the unequally weighted interference powers are estimates.

22. The apparatus according to claim 20, characterized in that the unequal weighting is based on powers of individual interferers.

23. The apparatus according to claim 22, characterized in that the powers of the individual interferers are estimates.

24. The apparatus according to claim 15, characterized in that the unequally weighted sum is also based on a fade rate.

25. The apparatus according to claim 14, characterized in that the plurality of received signals are received from a plurality of spatially diverse receiving elements.

26. The apparatus according to claim 14, characterized in that the plurality of received signals are received from a plurality of temporally diverse receiving elements.

27. An apparatus for processing a plurality of signals received from mobile wireless sources, characterized in that it comprises: an antenna array comprising a plurality of antenna elements each operatively connected to an antenna channel; a weight generating unit, operatively connected to the antenna array, for generating a plurality of weights for the antenna channels, based on a ratio of the desired signal power of a desired mobile user to a weightedly weighted sum of the power of the noise and the power of the interference; and a combination unit, operably connected to the weight generating unit, for weighting and combining the plurality of signals received using the plurality of weights generated to amplify the desired signal of the mobile user and suppress the interference signals.

The apparatus according to claim 27, characterized in that the unequally weighted sum is based on the accuracy of an interference estimate.

29. The apparatus according to claim 27, characterized in that the unequally weighted sum of the noise power and the interference power is based on an interference to noise ratio.

30. The apparatus according to claim 29, characterized in that the interference to noise ratio is an estimate.

31. The apparatus according to claim 29, characterized in that the interference to noise ratio is a known value.

32. The apparatus according to claim 29, characterized in that the relation is an estimate.

33. The apparatus according to claim 27, characterized in that the power of the interference comprises a plurality of unequally weighted interference powers.

34. The apparatus according to claim 33, characterized in that the unequally weighted interference powers are estimates.

35. The apparatus according to claim 33, characterized in that the unequal weighting is based on individual interference powers.

36. The apparatus according to claim 35, characterized in that the individual interference powers are estimates.

37. The apparatus according to claim 28, characterized in that the unequally weighted sum is also based on a fade rate.

38. The apparatus according to claim 27, characterized in that the plurality of received signals are received from a plurality of spatially diverse receiving elements.

39. The apparatus according to claim 27, characterized in that the plurality of received signals are received from a plurality of temporally diverse receiving elements.