MXPA98000853A - Des-extendedor adapta - Google Patents

Abstract

A method and an apparatus for adaptively de-extending an extended-spectrum direct-sequence signal. The direct sequence spread spectrum signal is provided to the cross filter (101). The output of the filter (101) is de-extended by a de-spreader (120) to provide a soft symbol determined by the symbol estimator (122). The soft symbol is provided to the decision circuit (124) which generates a corrected error version of the soft symbol data, referred to as a hard symbol. The hard symbol value is subtracted from the soft symbol by the subtractor (126) to generate an error symbol. The error symbol is then used to complete the derivation values of the transversal filter by means of an LMS algorithm used by the derivation adapter (10).

Description

ADAPTABLE DES-EXTENDER BACKGROUND OF THE INVENTION I_i. Field of the Invention The present invention relates to communication systems. More particularly, the present invention relates to a novel and improved apparatus and method for highlighting the performance of direct frequency spread spectrum receptors in the presence of unresolved multipath fading.

II. Description of Related Art Communications systems have been developed to allow the transmission of information signals from a source location to a physically different user destination. Both analog and digital methods have been employed for the transmission of these information signals over the communication channels linking the source and the user locations. Digital methods tend to provide several advantages over analog techniques, including, for example, better immunity to interference and channel noise, increased capacity and better communication security through the use of encryption. P1084 / 98 MX When transmitting an information signal from a source location on a communication channel, the information signal first becomes a suitable form for its efficient transmission on the channel. The conversion or modulation of the information signal involves varying a parameter of a carrier wave based on the information signal, such that the spectrum of the resulting modulated carrier is confined within the bandwidth of the channel. At the user's location, the original message signal is replicated from a version of the modulated carrier received subsequent to the preparation on the channel. This replication is usually achieved using an inverse of the modulation process employed by the transmitter. The modulation also facilitates multiple access, that is, the simultaneous transmission of several signals on the common channel. The multiple access communication systems will typically include a plurality of remote subscriber units that require intermittent service of relatively short duration instead of continuous access to the communication channel. The systems designed to allow communication in short periods of time with a set of subscriber units have been called multiple access communication systems. P1084 / 98 MX A particular type of multiple access communication system is known as the spread spectrum system. In the extended spectrum system, the modulation technique used results in an extension of the signal transmitted over a wide frequency band within the communication channel. One type of multiple access extension spectrum system is a code division multiple access modulation (CDMA) system. Other multiple access communication system techniques, for example, time division multiple access (TDMA), frequency division multiple access (FDMA), and AM modulation schemes, such as the simple amplitude sideband compressed and expanded. However, CDMA has considerable advantages over these modulation techniques for multiple access communication systems. The use of CDMA techniques in a multiple access communication system is disclosed in U.S. Patent No. 4,901,307, entitled "MULTIPLE SPECTRO ACCESS COMMUNICATION SYSTEM EXTENDED SATELLITE OR TERRESTRIAL REPEATERS", and the US Patent No. 4,901,307. United States No. 5,103,459, entitled "SYSTEM AND METHOD FOR GENERATING FORMS OF SIGNAL WAVE IN A CELLULAR TELEPHONE SYSTEM CDMA", both assigned to the assignee of this invention and P1084 / 98 MX are incorporated herein by reference. In a CDMA cell phone system, the same frequency band is used for communication in all cells. The CDMA waveform properties that provide processing gain are also used to discriminate between signals that occupy the same frequency band. further, high speed pseudonoise (PN) modulation allows these different spread transmissions to be separated, as long as the difference between the paths exceeds the PN chip length or bandwidth. If a PN chip rate of 1 MHz will be issued, the multipath demodulation can be used against paths that differ by more than one microsecond in the path delay from the desired path. A path delay differential of one microsecond responds to a differential path distance of 1,000 feet. The urban environment typically provides differential path delays greater than one microsecond and up to 10 to 20 microseconds are reported in some areas. In narrow band modulation systems, for example in the analog FM modulation employed by conventional cellular telephone systems, the existence of several trajectories may result in characteristics of P1084 / 98 MX severe fading. With broadband CDMA modulation, however, different trajectories can be discriminated in a demodulation process. This discrimination greatly reduces the severity of the ulti-trajectory fading. Multipath fading is not completely eliminated when using CDMA discrimination techniques, since there will occasionally be trajectories with delay differentials of less than the minimum path delay for the particular system. The signals that have path delays in this order can not be discriminated from each other in the modulator. Therefore, it is desired that there be some form of diversity that allows a system to be used in order to further reduce the effects of fading. The detrimental effects of fading can be controlled in some way by controlling the transmitter power in the CDMA system. The mobile unit and cell site power control system is disclosed in U.S. Patent No. 5,056,109, entitled "METHOD AND APPARATUS FOR CONTROLLING THE POWER OF TRANSMISSION PRODUCT IN A CDMA CELLULAR MOBILE TELEPHONE SYSTEM", assigned to the assignee of this invention and incorporated herein by reference. In addition, the effect of multipath fading can be reduced in the mode of P1084 / 98 MX communication transfer, when the mobile unit is going through a transition between the cell site service area to cellular sites that communicate to the mobile unit during the transfer process. The transfer scheme is disclosed in U.S. Patent No. 5,101,501, entitled "TRANSFER OF SOFT TRANSMISSION IN THE CDMA CELLULAR TELEPHONE SYSTEM", assigned to the assignee of the present invention and incorporated herein by reference. The existence of multiple trajectories can not provide a path diversity for a broadband CDMA system. If two or more paths with a differential path delay of more than one microsecond are available, two or more receivers can be used to separately receive the signals. Since these signals will typically exhibit independence in multipath fading, (ie, they will not normally fade together), the output of the two receivers can be combined in diversity. A method and apparatus for implementing a combination receiver of this kind is described in detail in US Patent NO. 5,109,390, entitled "DIVERSITY RECEIVER IN A CDMA CELLULAR TELEPHONE SYSTEM", assigned to the assignee of the present invention and incorporated herein by reference.

P1084 / 98 MX SUMMARY OF THE INVENTION The present invention is an improved and novel apparatus and method for improving the performance of the spread spectrum and direct frequency receivers in the presence of an unresolved multipath fading. The present invention is an alternative for the structure of the diversity receiver described in the aforementioned US Pat. No. 5,101,501, for direct frequency extension spectrum receptors. The present invention is similar in function to the diversity receiver, but has the advantage of being simpler and having an improved performance in the presence of unresolved multiple trajectories. It is also more appropriate for systems with high data rates that may be desirable in indoor applications. Its novelty lies in the incorporation of a de-extension and re-extension operation within a traditional adaptive equalizer. Multipath propagation channels that are in the indoor environment in the range of 800 to 2000 MHz typically have rather short delay extensions. The extension can vary from approximately 20 ns to 300 ns, depending on the size of the construction, the nature and design of the walls and other factors. The diversity receptors that are P1084 / 98 MX used for direct frequency reception in outdoor cellular environments are more effective when delays between multipath components are large compared to an extension sequence chip. In the standardized CDMA design, as described in detail in the aforementioned US Patent Nos. 4,901,307 and 5,103,459, the chip life is approximately 800 ns. The long chip duration in relation to the delay extension means that only a demodulated signal from the diversity receiver will be useful. In addition, the unresolved multiple trajectory will lead to a Rayleigh fading of the output that comes from that demodulated signal of the diversity receiver. In this way, the diversity receiver gain that is possible with longer delays is not achieved. The extension of short delay in interiors suggests that a novel method is needed for the management of the multipath signal. The present invention uses a structure similar to an equalizer to achieve this objective. The purpose of the present invention is to reduce intersymbol interference arising from multipath propagation. EQs that use the classic least squares (LMS) algorithm typically use P1084 / 98 MX feedback based on individual symbol decisions to update the derivation weights of a transversal filter. The LMS algorithm estimates the inverse gradient of an error function in relation to the derivation weights and adjusts them in a direction opposite to the estimated gradient. Under reasonable conditions of channel and gain statistics, the filter converges to a state that is effective in mitigating intersymbol interference. The LMS algorithm is widely used due to its simplicity, ease of computing and the fact that it does not need to repeat the data. However, in the present invention the LMS algorithm is not applied directly due to the direct frequency extension. In the similar CDMA and PCS system, the multipath extension introduces intersymbol interference not on the symbol time scale (tenths of a microsecond) but on the chip time scale (tenths to hundredths of a nanosecond). Therefore, the adaptive equalizer of the present invention works with feedback of chip-by-chip errors. In order to do this, the data modulation must be estimated from the unbalanced signal, the estimated error and the difference between soft and hard decisions re-extended by the original pseudonoise sequence, before P1084 / 98 MX feedback as a weight correction of the derivation. The exemplary implementation of the present invention is used to demodulate a pilot channel. A pilot channel is a channel used to provide basic timing tuning information and does not carry data. The use and implementation of a pilot channel is described in detail in the aforementioned U.S. Patent No. 5,103,459. The present invention can be used in the demodulation of other information channels with minor modifications.

BRIEF DESCRIPTION OF THE DRAWINGS The particularities, objects and advantages of the invention will be more evident from the detailed description set forth below, when taken in conjunction with the drawings in which the reference numbers are correspondingly conserved in all the drawings to identify like parts, and wherein: Figure 1 is a block diagram of the present invention.

DETAILED DESCRIPTION OF THE PREFERRED MODALITIES Figure 1 shows the structure of the adaptable and novel de-extender that allows the P1084 / 98 MX broadband channel equalization. A receiver that decreases the frequency of extension spectrum and converts it into a digital baseband signal is not illustrated in FIG., as is well known in the art. The adaptive de-spreader 100 is comprised of the cross-sectional filter 101, the average least squared derivative 103 update circuit (LMS) and the re-extension, de-extension and symbol estimation circuitry 121. The adaptive de-spreader 100 is illustrated constituted by discrete elements. In the embodiment of this example, the adaptive de-expander 100 is instrumented in a microprocessor or microcontroller programmed to perform the functions described. Although Figure 1 does not illustrate a separate timing block, it should be noted that this clock timing is typically provided within a microprocessor or microcontroller that could be provided by a separate timing source. The cross filter 101 is a finite impulse response (FIR) filter that receives the digital signal R, representative of a baseband version of the spread spectrum signal. The cross filter 101 generates a filtered signal based on the most recently received sample R (n), and the previously received samples R (n), R (n-2), R (n-3) and R (n-4) ). The structure P1084 / 98 MX shown in Figure 1 has five derivations. This is the filtered value, S (n) is based on five defined values R (n) and the previously received samples R (n), R (nl), R (n-2), R (n-3) and R (n-4) provided to the summing element 146. Although in the plain-axis mode the structure shown is a five-branch FIR filter, any number of derivations consistent with the performance requirements may be employed. The received signal is provided to a series of delay elements 102, 104, 106 and 108. In an extended spectrum communication system the smallest piece of information transmitted is referred to as a chip. Each chip has a predetermined chip period. Each of the delay elements 102, 104, 106 and 108 delays the received chip by a value equal to the related sampling period. In the example mode, the input sampling rate is the smallest integer multiple of the chip rate. For example, a chip rate of 1.25 MHz has an integer multiple which could be 4 with a corresponding sampling rate of 5 MHz. The delay elements 102, 104, 106 and 108 can be implemented using latches or other memory elements that are well known in the art. The current sample R (n) and the output of each of the delay elements 102, 104, 106 and 108 are P1084 / 98 MX respectively provide the multipliers 110, 112, 114, 116 and 118. In multipliers 110, 112, 114, 116 and 118, the sample values are weighted by weight derivation values where wi, v ~ 2 , W3, W4 and W5, respectively, which are computed by the LMS derivation update circuit 103, which will be discussed later. Each of the weighted sample values is provided to the adder (S) 146. The adder 146 adds each of the weighted sample values to generate the output value, Sn. The adder 146 outputs the filtered values Sn at the chip rate, ie one per chip period. The output chip of the transverse filter 101 Sn is provided to the de-extender, re-extension and symbol estimation circuitry, 121. The output chip, Sn, is provided to the multiplier 120. In the multiplier 120, the filtered chips of input Sn are multiplied by binary digits of a pseudorandom sequence provided by the pseudo-random frequency generator (PRS) 132, at the chip rate. The product of the multiplier 120 is provided to the symbol estimator 122. The symbol estimator 122 integrates the results of the outputs from the multiplier 120 over a period of symbols, which is greater than the chip period, in order to indicate the symbol P1084 / 98 de-extention MX. The symbol estimator 122 can be formed from a digital integrator well known in the art. The symbol estimator 122 may also provide a symbol map, wherein the integrated chip values provide a first symbol estimate that is applied to the soft symbol by a predetermined correlation. A typical correlation is the Hadamard transform correlation. If the extension is bipolar phase shift manipulation (BPSK), then the de-extender sequence should be evaluated as bipolar (± 1) identical to the extension sequence used in the transmitter. If the extension is manipulated by quadrature phase shift (QPSK) then generally its I and Q components must each be valued in bipolar form and the de-stretch sequence is the complex conjugate of the extension sequence used by the transmitter. In the exemplary embodiment, the PRS generator 132 is implemented using a displacement recorder, the design and implementation of which are known in the field. The de-strain operation provides a low-pass signal component corresponding to the data modulation present in the originally transmitted signal of interest. The interference from P1084 / 98 MX other stations will not be compressed by pseudo noise multiplication and will remain in a broadband wave area as described in detail in the aforementioned US Patent Nos. 4,901,307 and 5,103,459. The symbol estimator 122 operates at the output of the adder 146. The symbol estimator 122 filters or otherwise processes the samples that come from the transmitter filter 101 and generates an estimate of the modulation symbol that was transmitted during each symbol period. Normally, a large number of chips affect each symbol, reflecting the large proportion of the extension bandwidth with respect to the speed of the data. This output of the symbol estimator is marked as a "soft symbol" in Figure 1. The soft symbols are additionally produced by the resistance circuit 124. The output of the decision circuit 124 is a complex value that is a reconstruction of the modulation symbol. transmitted original. In the case of the demodulation of the pilot signal, this symbol is represented by a single complex number, for example 1 + 0j. In the case of the demodulation of a pilot signal, the decision circuit 124 is a comparator wherein the integrated chip values provided by the symbol estimator 122 P1084 / 98 MX are compared with a constant. This implementation of decision circuit 124 is useful for generating an unmodulated pilot estimate. On the other hand, a decision circuit 124 may be a complex circuit such as for example the Viterbi decoder that supplies re-encoded channel symbols, as a result of its final decisions. An error band form is calculated as the difference between the soft decision symbol provided by the symbol estimate 122 and a hard decision symbol provided by the decision circuit 124 in the adder 146. This complete error, e (n) , is re-extended in the multiplier 128 by a pseudorandom sequence provided by the PRS generator 132 which is delayed by the delay element 130. The delay element 130 can be configured from a memory or latching element known in the art. The updates of the derivation weight are computed according to the LMS algorithm in the LMS 103 update update circuit. The original extension signal samples should be used and not the modulation symbols. This differs from the traditional LMS adaptive equalizer in that the signal is de-extended for the elaboration of the symbol decision and is re-extended for the update of the derivation.

P1084 / 98 MX The structure of the decision circuit 124 may have different implementations depending on this application. In the exemplary embodiment, when a substantial portion of the uplink power is dedicated an extended but unmodulated pilot, the transmitted symbol is known a priori as constant. In this way, the output of the decision circuit 124 does not depend on the received signal in any way, but is simply a constant, such as 1 + jo. In this way, the soft symbols are short-term averages of the output of the des-extender and the symbol errors are differences between those short-term averages and the constant target. Symbol decisions are based on the values of many chips. Therefore, they are not available until a short time after the last chip from which they are included. Because of this delay, the received signal, R (n), must be delayed before being provided to the LMS derivation update circuit 103 and the random pseudorandom sequence must be delayed before being provided to the multiplier 128. The delay element 130 delays the pseudo-random sequence and the delay element 176 provides the delay for the received signal, R (n). These delays have a length of at least one data symbol. The delays will be P1084 / 98 MX if the lateral information of the Viterbi decoder is used since the final symbol decisions delay the reception in at least the truncation length of the decoder. The delays shown in Figure 1 are used so that the decision error feedback is in time alignment with the channel samples. Thus, they are approximately a symbol of duration. The typical symbol periods are 52.lμs. in the uplink and 208.3 μs in the reverse link. The extension code delay needs to remember only two bits per chip, or 128 and 512 bits. In this way, the signal samples, at a sampling of 8X and 4 bits for I and Q, would need 8192 and 32768 bits, respectively. In a system with a code rate p, modulation m-ava and data rate R, the symbol period is given by: _? .log2 m Tsimb ~ ~ '(!) so that the delay in terms of the samples is calculated as: f p.logr2 m - fahíp. s where S is the ratio of oversampling of the signal.

P1084 / 98 MX The number of derivations needed in the cross filter 101 can be calculated in terms of the total delay extension of the multipath signals. In this way, if the delay extension is set to 200 nanoseconds, then at this sampling rate of 20 MHz the number of derivations can be calculated as: 200 ns • 20 MHz + i > 5 derivations (3) that will be necessary. The success of this scheme depends on the speed of adaptation being fast enough to follow short-term changes in apparent multipath, mostly due to the movement of the handset. For a hand held unit, the rate of change at 1800 MHz and 3 m / s (6.7 mph) can be estimated to be 2fv / c = 36 zeros / second or approximately 28 ms between zeros. This suggests that the adaptation time would be no more than a few hundred microseconds for the scheme to be successful. At vehicle speeds the time is reduced by approximately 10 or approximately 2.8 ms. A fixed gain is shown by multiplying the error signal in the multiplier 134 before it is again provided to the derivation update circuit P1084 / 98 MX of the LMS 103. This gain must be selected properly, since it can cause the conversion to be slow if it is too much and can cause instability if it is too large. The LMS derivation update circuit 103 receives the weighted error signal from the multiplier 134 and the delayed samples from the delay element 176. The delayed samples from the delay element 176 are provided to a series of delay elements 168, 170, 172 and 174. The delay elements 168, 170, 172 and 174 each delay the sample received for a further sample period as described in the relation to the elements 102, 104, 106 and 108. The outputs of the delay elements 176 168 170, 172 and 174 is provided to multipliers 158, 160, 162, 164 and 166, respectively. The output of the multipliers 158, 160, 162, 164 and 166 provides a first input of the adders 148, 150, 152, 154 and 156, respectively. The outputs of the summing elements 148, 150, 152, 154 and 156 are provided to the delay elements 136, 138, 140, 142 and 144. The second input to the adders 148, 150, 152, 154 and 156 is an output delay of a single sample of each of the respective adders 136, 138, 140, 142 and 144. The elements of P1084 / 98 MX delay 136, 138, 140, 142 and 144 delay the input sample in a simple sampling period. The outputs of the delay elements 136, 138, 140, 142 and 144 are provided as the derivation values to the transverse filter 101 as provided to the multipliers 110, 112, 114, 116 and 118, respectively. This structure is simpler than that of the diversity receiver. Only one demodulator is required, in contrast to the various demodulators that are required in the diversity receivers. The need to search for multipath signals and assigning demodulation elements to multipath signals is also omitted, since the derivation locations are fixed at regular intervals. As there is no dynamic allocation, there is no loss due to assignment errors. As there is only one smooth decision output, there is no need for de-drift. At comparable levels of complexity more leads can be used, possibly leading to a better diversity gain. The prior description of the preferred embodiments is provided to enable any person skilled in the art to make or use the present invention. The different modifications to these modalities will be evident for those with expertise in this field, and the generic principles defined here P1084 / 98 MX may be applied to other modalities without the use of the inventive faculty. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but should be consistent with the broader scope consistent with the novel principles and features of the present invention.

P1084 / 98 MX

Claims (23)

1. NOVELTY OF THE INVENTION Having described the present invention, it is considered as a novelty and, therefore, the content of the following CLAIMS is claimed as property: 1. An adaptive de-extender comprising: a transverse filter means for receiving samples input and to filter the input samples according to a set of adaptive filter derivative values to provide filtered chip values, wherein the adaptive derivative values are updated according to a re-extended error signal; a de-extension means for receiving the filtered chip values and de-extending the filtered chip values according to a direct sequence extension spectrum format to provide a first estimated symbol and for generating a second estimated symbol according to a predetermined decision format; an error calculating means for receiving the first estimated signal and the second estimated signal and for generating an error signal according to the first estimated signal and the second estimated signal; an extension means to receive the error signal and to extend the spread spectrum of the error signal according to an extension spectrum format P1084 / 98 MX default, in order to provide the re-extended error signal. The apparatus according to claim 1, wherein the transverse filter means is a finite impulse response (FIR) filter. The apparatus according to claim 1, wherein the de-extension means comprises: a symbol estimator means for generating a first symbol estimate according to the filtering values; and a hard decision means that generates a hard symbol estimate according to the first symbol estimate, according to the predetermined hard decision format. 4. The apparatus according to claim 3, wherein the hard decision means comprises a comparator circuit. The apparatus according to claim 3, wherein the hard decision means comprises a Viterbi decoder. The apparatus according to claim 3, wherein the de-spreading means further comprises: a pseudo-random frequency generating means for generating a pseudo-random sequence: and a multiplier means for receiving and multiplying the filtered chip values and for receiving the sequences P1084 / 98 MX pseudo-randomly and to multiply the chip values filtered by the pseudo-random sequence, in order to provide a product sequence. The apparatus according to claim 1, wherein the transverse filter means updates the derivation values according to the average least squared derivation (LMS) adaptation format. The apparatus according to claim 6, wherein the symbol estimator serves to integrate the product sequence in order to provide a de-extension sequence. The apparatus according to claim 8, wherein the symbol estimator means further serves to correlate the de-extension sequence with a second sequence, according to a predetermined correlation format. 10. The apparatus according to claim 9, wherein the correlation format is a Hadamard transform. 11. A method for adaptively de-extending an extended spectrum signal, comprising the step of: receiving the input samples; filter the input samples according to a set of adaptive filter derivation values for P1084 / 98 MX provide filtered chip values, where the adaptive derivation values are updated according to a re-extended error signal; de-extending the filtered chip values according to a direct sequence extension spectrum format to provide a first estimated symbol; generate a second estimated symbol according to a predetermined decision format; generate an error signal according to the first estimated signal and the second estimated signal; and extending the error signal according to the predetermined extended spectrum format to provide a re-extended error signal. The method according to claim 11, wherein the filtering step comprises filtering the input samples in a finite impulse response (FIR) filter. The method according to claim 11, wherein the step of de-extending the filtered chip values comprises: generating a first symbol estimator according to the first filtered chip values; and generating a hard symbol estimate according to the first symbol estimate, with a predetermined hard decision format. 14. The method according to claim 13, in P1084 / 98 MX where the step of generating a hard symbol estimate comprises a comparison of the first symbol estimate with a set of threshold values. The method according to claim 13, wherein the step of generating a hard symbol estimate comprises the Viterbi decoding of the first symbol estimate. The method according to claim 13, wherein the de-extending step further comprises: generating a pseudo-random sequence; and multiplying the chip values filtered by the pseudorandom sequence to provide a product sequence. The method according to claim 11, further comprising updating the derivation values according to at least one least-squares average derivation (LMS) adaptation format. 18. The method according to claim 16, wherein the de-extending step further comprises integrating the product sequence to provide a de-spiked sequence. The method according to claim 18, wherein the de-extending step further comprises correlating the de-extended sequence in a second sequence according to a correlation format P1084 / 98 MX default. 20. The method according to claim 19, wherein the correlation format is a Hadamard transform. 21. An adaptive de-extender comprising: a transverse filter having a first input for receiving input samples and having a second input for receiving derivation update values and having an output; the error calculator has an input coupled to the output of the transverse filter and having an output to provide a calculated spread spectrum error signal; Y; a derivation update calculator having a first input coupled to the output of the error calculator and having an output coupled to the second transverse filter input. 22. The apparatus according to claim 21, wherein the error calculator comprises: a symbol estimator having an input and an output; a decision circuit having an input coupled to the output of the symbol estimator; and a subtractor having a first input coupled to the output of the symbol estimator and a second one P1084 / 98 MX input coupled to the output of the decision circuit and having an output. 23. The apparatus according to claim 22, wherein the error calculator further comprises a de-expander having an input coupled to the transverse filter output coupled to the input of the symbol estimator. P1084 / 98 MX