HRP921479A2 - Method and apparatus for diversity reception of time-dispersed signals - Google Patents

Description

This invention relates generally to diversity receivers in communication systems and more specifically to diversity receivers for time-distributed signals in communication systems.

Improved signal detection in a time-distributed medium generally requires the receiver to perform some type of equalization of the reflected signal with the received time-distributed signals to produce an output that is a better output than if the reflected signals were allowed to interfere with each other. One such equalization technique using a time division digital radio system (TDMA) is described in US Patent No. 4,829,534 entitled "Phase Coherent TDMA Quadrature Receiver for Multipath Fading Channels" of the current assignee filed in the name of Borth et al.

The invention by Borth et al describes a phase coherent method for quadrature phase keying (QPSK) demodulation of a radio signal subjected to multiple attenuation. equalization is enabled by correlating memorized preparatory sequences, known to the receiver, with the incoming signal, and using the resulting correlation to drive the phase difference between the incoming signal and the receiver's local oscillator, using coherent detection. The equalization can then be continued. Another technique is proposed for working with intersymbol interference that can be generated in a transmission signal over a time-distributable transmission channel. Such receivers are described in "Adaptive Maximum Likelihood Receiver for Carrier-Modulated Data-Transmission System", published by G. Ungerboeck, IEEE Transaction on Communications, Vol. COM-22, No. 5, May 1974, pp. 624-636 and "Maximum Likelihood Sequence Estimation of Digital Sequences in the Presence of Intersymbol Interference" by G. D. Forney, IEEE Transactions on Information Theory, Vol. IT-18 No. 3, May 1972, pp. 363-377.

However, in high-data-rate systems where the transmission is over a radio channel with a large delay spread, a single-ended equalization receiver can provide an adequately spaced (multi-ray attenuation) distortion correction. For example, a practical equalization receiver installation may have an incomplete estimate of the critical signal error in the case of a feedback equalization solution, or an incomplete estimate of the channel impulse response transfer in some other equalization scheme.

Therefore, diversity reception (of some signal recorded on multiple branches - which can be on different antennas or on the same antenna at different times, or done in a different way, as is known in the art) is typically insufficient to sufficiently reduce the effect of multi-ray attenuation. One such patent is described in US Patent No. 4,271,525 entitled "Adaptive Diversity Receiver For Digital Communications". That patent describes an adaptive diversity receiver that uses an adaptive transverse filter for each receive branch followed by a feedback decision equalizer. These transversal filter input outputs are improved via feedback via the equalizer output, and other points in the receiver.

US Patent 4,731,801 entitled "Method For The Reception And Detection Of Digital Signals" discloses an improvement over US Patent No. 4,271,527 and other prior art by improving reception in highly dispersive transmission paths using coherent demodulation. This invention uses the technique within the output of the switching technique for bits and becomes the basis for calculating signal correction. The reference carrier frequency obtained by summing the modulation frequency quadrature band signal and the modulation frequency band signals that are in phase is fed back into the local oscillator of the quadrature demodulator which backward compensates for the phase difference between the received signal and the receiver's local oscillator to enable coherent demodulation.

However, the invention as described in US Patent No. 4,271,525 calls for fitting custom transverse filters, one on each receive branch in addition to the equalization circuit. Inventions such as US Patent No. 4,731,801 require complex circuitry to phase-shift the signals in each of the various branches and, more importantly, cannot make phase-alignment corrections fast enough to be used, for example, in TDMA systems characterized by the received information having to be corrected at a short time separated by a relatively long time period. During this long period the phase of the signal in the multi-ray attenuation channel can change radically depending on the receiver's local oscillator.

Accordingly, there is a need here to reduce the complexity of a receiver that performs versatile reception on high-speed continuous or non-continuous digital signals and is capable of significantly reducing the effects of both wide fading and multi-ray, dispersive attenuation specific to a time-distributable time medium.

These requirements are substantially satisfied through the methods and devices for the diverse reception of time-distributed signals in the communication systems described below. The described procedure includes the correlation of the first time-distributed signal received on the first receiving branch with a known reference giving the first correlation signal, and the correlation of the second time-distributed signal received on at least the second receiving branch with a known reference giving the second correlation signal, that using the correlation signals the presetting of the first time-distributed signal and a second time-distributed signal for a known receive signal reference and a local branch oscillator providing the first aligned signal and the second aligned signal and generating the resulting signal in the form of the first aligned signal and the second aligned signal.

The known reference signal is located in a memorized lookup table (which contains multiple synchronization sequences as provided in the case of either a time distribution /TDMA/ system or a frequency distribution /FDMA/ system with built-in reference signals). Correlation determines, among other things, the prediction of the impulse response of a radio transmission channel. After completing the correlation, a tuned filter, usually a transverse filter having derivatives for the estimated channel response, is used to perform convolution on the time-distributed received signals, by performing phase equalization. Phase equalization essentially compensates for the phase difference between the received time-distributed signal and the local oscillator in each receive branch.

Samples of the processed signal obtained from each branch are selected, with additional parameters related to the channel impulse response (s-parameters) in the sample selection technique and then all fed into the sequence estimator to complete the equalization process on the newly formed signal. Various signal combining techniques, such as the well known maximum level combining or equal input combining techniques can also serve as acceptable diversion techniques for combined phase combined signals and channel pulse dependent parameters that can be used for equalization.

The invention is explained in more detail with the help of the attached drawing on which:

Figure 1 shows a block diagram generally describing the invention;

Figure 2 shows a more detailed block diagram of the invention where quadrature reception of digital signals is used;

Figure 3 shows a block diagram generally describing a diversity processor using bit-by-bit diversity selection according to an embodiment of the invention;

Figure 4 shows a block diagram that generally describes a diverse processor to which the maximum level combining process according to the invention is applied;

Figure 5 shows a block diagram that generally describes the second maximum level combining process according to the invention, i

Figure 6 shows a block diagram describing the invention using adaptive linear equalization.

Figure 1 generally depicts the invention as an adaptation of the equalizing receiver described by G. Ungerboeck and explained above. Quadrature phase keying (QPSK) demodulation was applied. However, other digital multidimensional signaling, such as GMSK, can also be used. As shown, the invention includes a first receiving branch 100 and at least a second receiving branch 101, each branch comprising a state of radio frequency demodulation and only part of the processing necessary to correct the time-distributed received signal 102 and 103. Although the description of the invention speaks of a receiver having two branches the disclosed invention is easily applied to receivers having N branches.

Both receive branches are provided with a stored expected reference sequence 104 to enable post-demodulation correlation between the received time-distributed signals and the reference sequence using a known correlation technique. Correlation information provides synchronization information, and parameters from which channel the tuned filter can be made. The received time distributed signal is processed through this tuned filter and the resulting phase tuned signal from each branch is then used via the diversion determination block 105 to generate the output signal. The correlation further provides information that can be used to predict input controls 106 to maintain the received signal within a specified voltage range. Not renewing the circuit of the carrier frequency (phase compensation feedback) is necessary for correct modulated time-distributed signals in relation to the local oscillator in each branch ensuring that the response of the channel pulse transmission, according to the correlation and according to the outputs of the tuned filter, does not change significantly during the time period when the data to be processed is received. In any TDMA or FDMA system, it may be necessary to adjust the local oscillators on the receiver branches or to process the received data through another means, if this is not the case.

A more detailed block diagram of the invention is shown in Figure 2. The first diversity receiver branch 100 and the second diversity receiver branch 101 receive signals transmitted from the same point, but which will become time-distributed, in different ways, en route from the transmitter to each receiver branch. These signals are processed by a known technique and are amplified and mixed in the intermediate frequency states 200 and 201. After that, the signals are demodulated (transformed for the range of the modulation frequency) into the in-phase signals, 11 and 12, and the quadrature phase signals Q1 and Q2 via the quadrature demodulators 202 and 203, known in the art, whose inputs come from local oscillators 208 and 209. It is already known in the art that the same local oscillator can be used for different branches.

Each of the in-phase and quadrature signals of each branch is passed to its broadband filter 220, 230, 240, 250 and examined by an analog-to-digital (A/D) converter 204, 205, 206, 207. Each of these signal samples brought is in the correlators so that the tested signal 11 is fed into the correlator No. 1 (210), the tested signal 12 is fed into correlator No. 2 (212), Q1 is also fed into correlator No. 1 (210) and Q2 also fed into correlator No. 2 (212). The known reference 104 is stored in correlator memory or some other convenient location, as a reproduction of the expected bit or pattern, to capture amplitude and phase information substantially simpler than an ideal signal not subjected to time-distributed attenuation.

The selected signals are correlated with the memorized reference information which is given in several sizes. This involves estimating the impulse response of the transmission link (or measuring the time-distributed distortion of the medium), then estimating the frequency/phase offset relative to a known reference, and some measurement of the signal strength. The power measurements are from the input to the input controller 106 which in turn individually adjusts the input of each branch according to a predetermined equalization input suitable for the application. As is already known in the art, the input of each IF (intermediate frequency) state 200 and 201 of each branch 100 and 101 can be simultaneously adjusted to produce the same input on each branch.

Other sizes were used to construct the channel-tuned filter for each branch, as is known in the art. After passing the sampled signal through each tuned branch filter 214, 215, the output signals have the effects of time-distributed distortion and phase error with each local branch oscillator having shifted significantly.

For example, in a radio TDMA communication system, a synchronization sequence is used in one allocated time frame of a transmitted short pulse, the memorized reproduction of the synchronization sequence expected in the receiver can be the data memorized in the lookup table. The receiver correlates the received signal of both branches 100 and 101 with the memorized expected sample synchronization to determine the time-distributed model profile (channel impulse response) of each branch calculates the adjusted filter coefficients based on the samples obtained by correlation or prediction of the channel impulse response.

The phase-aligned signals outside the tuned filter 214 in the first branch 100 are displayed via IM1, QM1 while the phase-aligned signals outside the tuned filter 215 in the second branch 101 are displayed via IM2 and QM2. The diversity processor 105 then uses at least a proportion of at least one of these aligned signals from both branches (and other relevant information, such as signal strength measurements) to best determine the data of the original transmitted signal.

The diversity processor 105 can use different techniques to achieve optimum rendering of the original transmitted signal.

Figure 3 describes a diversity processor that uses bit-by-bit diversity selection within selected branch correlation parameters, called s-parameters, and selected samples from the aligned signals that attack a sequential equalizer that generates a representation of the original transmitted signal.

A more detailed description of the generation of s-parameters in the receiver is given in the Ungerboeck report, cited above, by equation 17. Each branch derives its own s-parameters for correlation, based on the convolution of the estimated channel impulse response and the impulse response of the respectively tuned filter.

The aligned signal of the first branch AS1 and the aligned signal of the second branch AS2 are input to the processing state 300. One sample for the transmitted data symbol of each of the aligned signals is compared to the corresponding transmitted sample of the data symbol of each branch. The absolute values of these samples are compared and the current sample with the highest absolute value is placed in a group of symbols that will later pass through the sequence evaluation state 305, which comprises sequence evaluation as is known in the art. Although the highest absolute value is the basis for selection in this embodiment, the lowest absolute value or other suitable basis may also be used.

Moreover, a collector for each branch is available to record the number of samples selected from each branch so that they are placed in a group of symbols. When the last symbol of the signal patterns is compared, the counters are compared to determine which branch provides the most samples for the symbol group. The s-parameters (ss) from the branch providing the most samples for a group of symbols are sent to the sequence estimator with certain intersymbol interference information. The sequential estimator then completes the equalization process.

The s-parameter selection can be eliminated and a set of s-parameters chosen randomly from one of the branches can be used in the Ungerboeck receiver, but better characteristics are desired. A receiver using a frequency estimator is not of the Ungerboeck type described and does not use s-parameters, but still represents a selection based on the symbol patterns following the tuned filter and represents an equal selection from any distortion and compensation channel parameters that may be used via the sequence estimator.

As mentioned combining the appearance of signals in the middle of the equalizer to facilitate diversity, inasmuch as an equalizer of this type in one receiver branch can be considered a combination of a tuned filter and a sequential estimator. The present invention only wants to unify the function of the tuned filter, but not the function of the sequential estimator in the diversity receiver.

Figure 4 describes another diversity processing technique, instead of using combining analogously to the maximum level of combining aligned signals, where the weighting factor is determined by the strength of the received signals according to the technique described in the patent application of the current assignee of rights No. 07/358325, filed on May 26, 1989. year, under the name "Rapid Received Signal Strength Indication" found by Labedz and others. The gravity factor is best derived from the summation of the square of the quadrature components of the energy at the relative maximum of correlations between the measured reflected signals and the memorized reference sequences. Such correlated energy measurements are integrated to determine the energy displayed between multiple time-distributed reflected signals, and the resulting key factor is called the "Received Signal Strength Indicator". However, a sample or integration of several samples of the received signal envelope may also be used.

The received signal strength indicator for the first branch RSSI1 400 is multiplied by the aligned signal from the first branch AS1 using a multiplier 410 to form the key aligned signal for the first branch. The received signal strength indicator for the second branch RSSI2 405 is multiplied by the aligned signal of the second branch AS2 using the multiplier 415 to form the key aligned signal for the second branch. These key signals are then summed at point 420 to provide a signal comprised of the key signals from both branches.

The S-parameters described above are processed in a similar way. The received signal strength indicator for the first branch RSSI1 400 is multiplied by the s-parameters from the first branch (s-para1) using the multiplier 430 and forming the key group of s-parameters from the first branch. The received signal strength indicator for the second branch RSSI2 405 is multiplied by the s-parameters from the second branch (s-para2) using the multiplier 440 and forming the key group of the s-parameters from the second branch. These key parameters are then summed at point 450 to provide a signal comprised of the key group of s-parameters from both branches. This technique can also be used when using N receiving branches.

Again, combining the s-parameters can be eliminated in the Ungerboeck receiver, but better characteristics would be obtained this way. A receiver using a sequence estimator is not the type described by Ungerboeck that uses s-parameters, but still performs combination based on the symbol pattern following a tuned filter and the intended combination of any distortion-compensation parameter can be used via the sequence estimator.

In the case where no signal strength weights are used, if RSSI1 400 and RSSI2 415 are effective groups equal to 1, a similar technique to obtain an equal input to combine gives and multiplier 410 and 415 and 430 and 440 are no longer necessary.

Figure 5 describes another method of variously incorporating a suitable technique for the maximum level of combining aligned signals using a signal strength indicator (SSI) obtained from determining the signal strength measured at the intermediate frequency states 200 and 201 of each branch. This variant combines the complex (in-phase and quadrature-phase) aligned signals from each branch, before they pass through the complex signal demultiplexer 520 .

Weighted signal strength equalization 500 determines the relative weight assigned to the in-phase aligned signal (IM1 and IM2) in each branch and the quadrature-phase aligned signal (QM1 and QM2) in each branch. This equalizes the weights of IM1 and QM1 via RSSI1/(RSSI1 +RSSI2) and the weights of IM2 and QM2 via RSSI1/(RSSI1+RSSI2). The key input signals (ISS1 and ISS2) are summed at point 510 giving a combined quadrature signal for both channels. Equal input of combining signals or bits inside can also be performed by a suitable diversity decision technique. Such an equal combining input for diversity reception would again involve setting the values of RSSI1 and RSSI2 equal to 1.

Although the illustrated embodiment is suitable for comparison in systems with high speed, non-continuous signals as in TDMA systems having a short signal interval, an alternative embodiment of the invention may be envisaged when receiving sufficiently long data streams where the transmission of the channel impulse response changes significantly over time. period, during which the processed data was received.

Figure 6 shows one branch of the present invention utilizing an adaptive linear equalizer. In this variant, the initial correlation is performed using a known reference to estimate the channel impulse response (CIR) and the feed line equalizer (Ck(0)). The output of the equalizer feed is then adjusted using a typical adaptive linear equalization technique (as described in "Digital Communications" by John G. Proakis, McGraw-Hill Book Company 1983, pages 357-386), to continue pre-tuning the received signal. Therefore, coherent pre-tuning of the received signal involves at least correlating the signal with a known reference (initial) and then later, adjusting the input terminal of the equalizer in accordance with the known technique of adaptive linear equalization.

As shown, the received signal passes through the state quadrature demodulator 600 and then after selection and digitized in the state A/D converter 605 which gives a complex signal (R(n)). This signal is correlated with a stored reference in a correlation state 210 resulting in a channel impulse response (CIR) used in the initial feed terminal equalizer (Ck(n)) calculation 615 and signal amplitude estimation 620. The signal amplitude estimation criteria contribute to the key factor being determined in the weighted state of the signal 265. As disclosed, R(n) is also the input to the equalizer 630 within which the soft information (S(n)) is generated, in a known manner. The soft information (S(n)) at the output of the equalizer is directed to a decision circuit 635 within which a tentative decision S(n) is made in terms of improving the equalizer output 640 feed, as a received signal. A suitable S(n) signal is generated, which is keyed in block 625 and output to a summing point 645 which combines some or all of the outputs of this branch (in the case of a two-branch receiver) before the final bit decision. The output signal of each branch may be provided mainly to combine the other equalization feed or the criterion of the maximum level of combining or may be combined using bit-by-bit selection, as previously disclosed.

As those of ordinary skill in the art know, the present invention can also be applied to receivers using decision equalizers in a feedback loop or any non-linear equalizer provided for example, S(n) can be soft information derived from a decision equalizer in the feedback loop before the decision loop. Also, symbol pattern selection, or combining, may occur after the pre-feed filter (which is also known in the art) for each diversity branch or any other intended point within the equalization block itself.

Claims (21)

1. A receiver for providing diversity reception having correlation means 210, 212 for generating first and second correlation signals from first and second time-distributed signals received on first and second reception branches 100 and 101, redistributing means 214, 215 operatively connected to the correlation means 210 , 212 for generating a first feedback signal using a coherent preset first time-distributed signal with a reference frequency through the use of at least a first correlation signal and for generating a second feedback signal through a coherent second time-distributed signal with a reference frequency using at least a second correlation signal and signal generation means 105 operative connected to redistributing means 214, 215 for generating a signal derived from the first and second aligned signals, has an improvement, characterized in that the correlative means (210, 212) characterized by the means for generating the first correlation og signal through the correlation of the first time-distributed signal received on the first receiving branch (100) with a known reference (104) and means for generating a second correlation signal through the correlation of the second time-distributed signal received on the second receiving branch (101) with a known reference (104). 2. Receiver according to claim 1, characterized in that the known reference (104) within it is characterized by a known synchronization sequence. 3. The receiver according to claim 1, characterized in that the means for generating the signal (105) within it are characterized by means for ensuring bit-by-bit diversity selection (300, 305). 4. The receiver according to claim 1, characterized in that the signal generating means (105) within it are characterized by means for ensuring a maximum combining ratio (420, 450). 5. Receiver according to claim 2, characterized in that the known synchronization sequence within it has an assigned time frame of a short TDMA signal. 6. A receiver according to claim 3, characterized in that the correlative means (210, 212) therein are further characterized by means for generating s-parameters for both the first receive branch (100) and the second receive branch (101) based on the estimated response of the channel pulse on each receiving branch (100, 101) and the response of the tuned filter on each tuned branch. 7. The receiver according to claim 3, characterized in that the means (300, 305) within it for ensuring bit-by-bit diversity have: means (300) for selection between the symbol samples and the first aligned signal, so that this bit-by- bit selection gives a group of symbols from the selected symbol samples; means (300) for determining which aligned signal contributed to the predicted number of selected symbol patterns for the output group; and means (300) for providing the sequence estimator (305) with a group of symbol patterns and a group of s-parameters associated with the receiving branch whose aligned signal contributes to the predicted number of selected symbol patterns. 8. A receiver according to claim 1, characterized in that the means for generating the signal (905) has within it means for measuring the symbols of the signal samples from each aligned signal; symbol summing means operatively connected to the measurement means for summing the symbol patterns from each aligned signal to form a combined group of symbol patterns; parameter summing means operatively connected to symbol summing means, for summing the s-parameters associated with each receiving branch (100, 101) to form a combined group of s-parameters; and means, operatively connected to both summing means for providing the sequence estimator with the combined group of pattern symbols; and a combined group of s-parameters. 9. A receiver according to claim 1, characterized in that the signal generating means (105) within it are characterized by means for a maximum combining ratio where the key factor is determined by a received signal strength indicator (400, 405) determined for processing the channel by a sound technique using a time-distributable function, such as channel impulse response, to determine the energy level for receiving a time-distributed signal. 10. The receiver according to claim 1, characterized in that the means for generating the signal (105) within it are characterized by means for ensuring the maximum level of combining through the generation of a key factor based on the measurement of the signal strength measured in the intermediate frequency state. 11. A method for providing diversity reception, wherein the first and second correlation signals are generated from the first and second time-distributed signals received on the first and second reception branches (100, 101) by generating the first reception signal via a coherent preset correlated signal for the reference sequence using at least the first correlation signal, generating a second feedback signal via a coherent preset second correlation signal for the reference frequency using at least a second correlation signal and generating a signal derived from the first and second aligned signals has an improvement, characterized in that the generation of the first correlation signal via the correlation of the first time-distributed signal received at a first receiving branch (100) with a known reference (104); and generating a second correlation signal via correlation of the second time-distributed signal on the second receiving branch (101) with a known reference (104). 12. The method according to claim 11, characterized in that the known reference (104) has an expected known synchronization sequence within it. 13. The method according to claim 11, characterized in that the generation of the signal derived from the first and second aligned signals is characterized by providing bit-by-bit diversity selection. 14. The method according to claim 11, characterized in that the generation of the signal derived from the first and second aligned signals is characterized by ensuring the maximum level of combining. 15. The method according to claim 11, characterized in that a known synchronization sequence is assigned to the time frame of a short TDMA signal. 16. The method according to claim 11, characterized in that the generation of correlation signals is characterized by the generation of s-parameters for the first receiving branch (100) and the second receiving branch (101) based on the estimated channel impulse response of each receiving branch and the response of the adjusted filter for each receiving branch branch (100, 101). 17. The method according to claim 13, characterized in that ensuring bit-by-bit diversity selection is characterized by: selection between symbol samples in the first aligned signal and symbol samples in the second aligned signal, so that the bit-by-bit selection provides a group of symbols from selected symbol samples; determining which aligned signal contributed to the predicted number of selected symbol patterns for the symbol group; and by providing a frequency estimator with a group of symbols; and groups of s-parameters associated with the receiving branch whose aligned signal contributes to the predicted number of selected symbol patterns. 18. The method according to claim 11, characterized in that the generation of the signal derived from the first and second aligned signals is characterized by: measuring the symbol signal from each aligned signal; summing the symbol patterns from each aligned signal to form a combined group of symbol patterns; summing the s-parameters associated with each receiving branch (100, 101) to form a combined group of s-parameters; and providing a sequence estimator with a group of combined symbol patterns; and a group of combined s-parameters. 19. The method according to claim 11, characterized in that the generation of the signal derived from the first and second aligned signals is characterized by providing the maximum combining ratio where the weighting factor determined from the received signal strength indicator is determined by the processed sound channel technique using a time-distributed function, as a channel impulse response for determining the energy level of received time-distributed signals. 20. The method according to claim 11, characterized in that the signal generation derived from the first and second aligned signals is characterized by providing the maximum combining ratio through the generation of a key factor based on the measurement of the signal strength measured in the intermediate frequency state. 21. A method for providing diversity reception wherein first and second correlation signals are generated from first and second digitally time-resolved signals recorded on first and second receive branches (100, 101) by generating a first aligned signal via a coherent first correlative redistribution signal for a local oscillator (208) a first branch using at least a first correction signal, generating a second aligned signal with a coherent second correlative redistribution signal for a local oscillator (209) a second branch using at least a second correlation signal and generating a signal derived from the first and second aligned signals, characterized in that generating a first correlation signal by correlating the first time-distributed signal received in the first receiving branch with a known synchronization sequence (104); generating a second correlation signal by correlating the second time-distributed signal received on the second receiving branch with the known synchronization sequence (104); and generating a signal derived from the first and second aligned signals characterized by: sampling symbols from each aligned signal; selecting the intended symbol pattern from at least one of the aligned signals based on the absolute value of at least one symbol from each aligned signal such that a bit-by-bit selection is made between the symbols in the first aligned signal and the symbols in the second aligned signal and based on the absolute value of each pattern symbols so that it forms a symbol group of selected symbol patterns; by collecting the number of symbol samples selected from each aligned signal; and providing a sequence estimator with a group of symbols and a group of s-parameters associated with the receive branch (100, 101) having the desired number of summed symbol samples.