CA2068964A1

CA2068964A1 - Maximum likelihood sequence estimation based equalization within a mobile digital cellular receiver

Info

Publication number: CA2068964A1
Application number: CA002068964A
Authority: CA
Inventors: Michael I. Parr; John L. Watson; Thayamkulangara G. Vishwanath
Original assignee: Hughes Aircraft Co
Current assignee: Raytheon Co
Priority date: 1991-06-27
Filing date: 1992-05-19
Publication date: 1992-12-28
Also published as: US5537419A; ATE165702T1; AU650206B2; EP0520969A3; NO922363L; JPH05327785A; TW201372B; EP0520969B1; US5563888A; AU650593B2; US5263026A; US5471501A; FI922388A; JPH0828751B2; AU1854592A; DE69225274D1; AU643147B2; EP0520969A2; DE69225274T2; NO922363D0

Abstract

MAXIMUM LIKELIHOOD SEQUENCE ESTIMATION
BASED EQUALIZATION WITHIN A MOBILE
DIGITAL CELLULAR RECEIVER
ABSTRACT
The present invention provides for an equalization method that forms the nucleus of a receiver for digital cellular mobile telephones. A number of distinct aspects of the design are considered novel, including non-real time operation that permits and exploits "time-reversed" equalization to significantly enhance bit error rate performance. The present invention employs the use of a maximum likelihood sequence estimation process. It uses variable coefficient least mean square tracking during estimation of the transmission channel's impulse response. It also employs the use of integrated symbol timing adjustment and carrier tracking algorithms. Time-reversed operation further enhances performance and permits implementation with reasonable complexity of a standards compliant mobile receiver.

Description

2~9~

MAXIMUM LIKELIHOOD SEQIJENCE ESTIMATION
BASED EQUALIZ~TION WITHIN A MOBILE
DIGITAL CELLULAR RECEI~EPc BACKGRO{JND
The present invention relates generally to digital cellular cornmunications, andmore particularly, to a maximum likelihood sequence estimation based equalization method for use in mobile digital cellular receivers.
Communication channels in the cellular enviromnent cornmonly impose a com-5 bination of distorting effects on transmitted signals. Rayleigh fading, where a signal'sperceived power level rises and falls rapidly over a wide range, results from the combi-nation of signals that have traversed paths differing in length by at least a significant fraction of a wavelength (i.e., about 30 cm. for cellular). Differences in path transmis-sion times that approach the time taken to transrnit a symbol result in a second problem 10 called delay spread.
Delay spread results in reception of multiple delayed replicas of a transmi~ted signal. Each Rayleigh faded replica has randomly distributed amplitude and phase, and the rate at which this complex quantity varies is constr~ined by the Doppler bandwidth associated with a vehicle's speed In a ~equency nonselective environment, the 15 sampled outputs of a receiver's matched filter provides uncorrelated estimales of the transmitted data. As such, in terms of discrete time sampl s, the channel has exhibited an irnpulse response proportional to a delta function. With delay spread, on the other hand, the discrete time channel impulse response is extended to introduce energy at a number of symbol times. The effect of the channel on the ~ansmitted signal, in turn, 20 may be viewed as the convolution of the transmitted information with the channel's impulse response. The channel, therefore, emulates a convolutional coding p}~cess.

, , .

2 ~ 9~

This leads to the possibility of estimating the transmitted information through the use of methods analogous to typical decoding of convolutional codes, i.e.
maximum likelihood sequence estimation techniques. Unlike the more widely applied forward e~ror correction decoding environment, the details of the encoding process in a reverse error correction decoding environment, are not known a prion by the receiver. Issues related to the need to estimate the form of the encoding process are addressed by this invention.
It would therefore be desirable to provide an enhancement to the processing performed by equalizers for use in mobile telephones that provides for system complexity reduction and that provides for better performance in a fading channel.

SUl~IMARY OF T~IE INVENTION
The present invention provides for an equalization method that can form the nucleus of a receiver for digital cellular mobile telephones. It provides many advantages1 including: a non-real time operation mode that permits and exploits "time-reversed" equalization, which signi~lcantly enhances bit error rate performance;
the use of maximum likelihood sequence estimation; the use of variable coeffisient least mean square tracking during estimation of the transmission channel's impulse response; and integrated symbol timing adjustment and sarrier tracking algorithms.
More particularly, the present invention comprises a method of processing received symbols including known preamble data and transmitted data, which method compensates for the effects of a power ~fade caused by a fre~uency selective fading channel. One specific aspect of the present invention comprises a method of processing samples received from a delay-spread, fading channel, which samples are associated ~,vith a block of data transmitted within a time slot, and wherein the method is adapted to make data decisions, and comprises the following steps~ (1)Storing samples receiving during the time slot; (2) ~stimating the location within the time slow at which decision errors are most probable, by determining the location in time of the maximum fade depth in the transmission channel impulse response; (3)Processing the stored samples, starting with the first received sample and proceeding beyond the location of 2 ~ 6 ~

the maxirnum fade depth, using a predeterrnined maximum likelihood sequence estima-tion procedure to generate estimates of the transrnitted data; (4) Processing the stored samples, starting with the final received sample and proceeding in a reverse direction with respect to the time sequence in which the sarnples were stored, beyond the above-S deterrnined location, using the maximum likelihood sequence estimation process togenerate estimates of the transrnitted data; (5) Simultaneous with the above two pro-cessing steps, generating estimates of the characteristics of the transmission channel impulse response, which are used in the maximum likelihood sequence estimation process; (6) Processing the output of the preceding estimate generating steps to10 generate data decisions.
The step cf generating the channel impulse response estimates ~pically com-prises using variable tap coefficients that are determined by estimating tap settings within the estimated channel impulse response by minimizing the square of the differ-ence between actual received sarnples and those synthesized by passing known trans-15 rnitted signals through the estimated channel. The processing is done in an iterativemanner by combining previous estimates of channel impulse response and new esti-mates thereof based on recent information, and by varying the ratio of the contributions from the previous and new estimates as a function of location within the tirne slot.
The method may further comprise a symbol tirning adjustment procedure com-20 prising the following steps: (1) Using a subset of the samples received during a timeslot, generating an error measurement comprising a measure of the degree to which the estimated channel impulse response matches the actual channel impulse response; (2) Generating a plurality of similar measures utilizing simultaneously recorded samples having dif~erent time offsets where;n at least one sample is advanced and one sample is 25 retarded in time relative to the above measure; (3) Searching for a bit time setting that minimizes the above error measurement and adjusting the sampling to reflect the newly determined bit time setting.
The method may also further comprise a carrier offset ~racking method compns-ing the following steps: (1) Recording at least two samples of at leas~ one tap within the 30 estimated channel impulse response at selected symbol locations within the time slot during the equalization process; (2) Generating frequency offset estimates at each of a plurality of time slots by observing the phase difference in each time slot between the at least two samples; (3) Combining this plurality of frequency offset estimates by using a filtering process to generate a precise frequency offset estimate; (4) Adjusting a control-35 lable frequency source to compensate for the precise ~equency offset estirnate.
Also, the present invention also uses a foIward maximum likelihood es~imationmethod of processing samples received from a delay-spread, fading channel,which is 4 2 ~

adapted to make data decisions. This method comprises the ~ollowing steps: Storing samples received during the time slot; Processing the stored samples, starling wilh Ihc firsl received sample and proceeding beyond the last received sample of the time slot using a predetermined maximum likelihood sequence estimation procedure to generate estimates of the transmitted data; Simultaneous with lhe above processing, generating estimates of the characteristics of the transmission channel impulse response which are used in the maximum likelihood sequence estimation process; Processing the outputs of the preceding steps to generate data decisions.
The maximum likelihood sequence estimation method offers significant performance advantages when compared with alternative equalization options, such as decision feedback equalization, for example.
With respect to meeting industry defined standards for performance of digital cellular telephones used in fading environrnents, the perf~rmance of the equalizer incorporating the maximum likelihood sequence estimation process of the present invention stands out as superior. It is the only known process that is capable of meeting, or even approaching, the current digital celhllar mobile telephone specifications. Time reversed operation further enhances performance and permits implementation with reasonable complexity of a standards compliant mobile receiver.

BRIEF DESCRIPTION OF THE DRAWINGS
The various features and advantages of the present invention may be more readily understood with reference to the following detailed description taken in conjunction with the accompanying drawings, wherein like reference numerals designate like structural elements, and in which:
FIG. 1 is a graph of a fading channel's amplitude with respect to time and a diagram of the signals in that channel;
FlG. 2 is a block diagram of a digital cellular mobile telephone receiver incorporating a maximum likelihood sequence estimation based equalizer in accordance with the principles of the present invention;
FlG. 3 shows the processing performed in the maximum likelihood sequence estimation based equalizer of FIG. 2;
FIG. 4 is a graph of relative error on the vertical axis versus bit timing offset on the horizontal axis for a typical received signal of the present invention;
FIG. 5 is a flow diagram illustrating the processing performed by the equalizer of the present invention to implement carrier frequency offset compensation; and FIGS. 6A and 6B are flow diagrams illustrating the processing per~formed by the equalizer of the present invention to implement bit timing.

DETAILED DESCRIPI'ION
Referring to the drawing figures, FIG. 1 illustrates a problem associated with reception in a mobile environment having a fading channel. FIG. 1 is a graph showing received power level from a typical fading channel on the vertical axis versus time on the horizontal a~is. The location of the power fade F is shown relative to a lypical time slot. The diagram below the graph sko vs the frame structure of the fading channel versus the same time axis. Below the frame structure, the time slot is shown enlarged and includes a preamble PR and a coded digital verification color code field (CDVCC3, which 2 ~

comprise known data that is used to initialize a receiver systcm employing ~he equalizer of the present invention. This is followed by a postamble PO which is the preamble for the next symbol of the channel.
At the lowest portion of FIG. 1, equalization processing in accordance with the preseDt inYentiOn is illustrated with the arrows nAn and "B", in which the equaGzer of the present inven~ion processes ~or~vard and time reversed computations through the location of the power fade in order to accomplish the objectives of the present invention. This will be more fully described below with reference to FIGS. 2 and 3.
FIG. 2 is a block diagrarn of a digital cellular mobile telephone receiver system 20 incorporating a maxirnum likeGhood sequence estimation based equalizer 21 in accordance with the principles of the present invention. The system 20 comprises an amplifier 22 whose output is coupled by way of a downconverter, comprising a frequency source 23 and a mixer 24, to an analog filter 25. An analog to digi~al converter 26 is coupled to the analog filter 25 ;D order to digitize the downconverted data. A
matched filter 27 is coupled bet veen the analog to digital converter 26 and the equalizer 21 of the present invention. The equalizer 21 comprises a memory 30, a 4-state equalization trellis 31 that is adapted to calculate maximum likelihood sequence estimation metrics~ a channel impulse response estimator 32, and an equalizer control circuit 33.
A serially coupled AGC circuit 35 and gain control circuit 38 are coupled to the amplifier æ.
The equalizer control circuit 33 is coupled to an output of the matched filters ~7 and is coupled to an input to the &equency source 23. Symbol sampling (bit timing) time control circuitry 37 is coupled to the equalizer control circuit 33 and the acquisition circuit 36 and provides control signals to the analog to digital converter 26. The output of the matched filters 2~ is coupled to the AGC circuit 35 and the acquisition circuit 36 and to the equalizer control circuit 33 that is employed to control the frequency source 23 and provide training data for use in initiali~ing the equalizer 21.
In operation, a partially filtered IF signal with a center frequency of 85.05 MHz enters the gain controllable amplifier 22. The resulting sigr~al is then downconverted using the frequency source 23 and the mixer 24 to 461.7 kHz This signal is then filtered using a narrow analog filter 25 to reject most of the received signals outside the 30 kHz band of interest. The resulting signal is then sampled and converted to 8-bii digital samples using the analog to digital tA/D) converter 26. A 16 tap fractionally spaced digital FIR filter 27 then performs matched filtering to produce symbol spaced samples which enter the equalizer 21. Temporally offset matched filters 34 that are substantially the same as the matched filters 27 are provided for use by the syrnbol timing control circuit 37, via the equali~er control circuit 33.
The basic principles of maximum likelihood sequence estil'nation are well kDown and are based on Viterbi decoding. The maximum likelihood sequence estimation process eall be outlined as follows.
The channel has an impulse response containing significant energy in, say, N symbols. Assume that the transmitter sends a sequence of symbols, much longer than N. The transmitted sequence may be described as the transitions betveen states, where each state corresponds to a group of N-1 transmit~ed syrnbols. The states, therefore, correspond to overlapping groups o~ transmitted symbols. In consecutive states, therefore, all but one constituent symbol are the same, and the possible transitions between states are correspondingly constrained. As each sample is received, ~he equalization trellis 31 considers eve~y possible sequence of N symbols that could have contributed to its value, by convo}ving that sequence with 2 ~

the estimated channel impulse response. For each hypothesized sequence, the result of the convolution corresponds, or fails to correspond, in some way (defined by a statistic called a metric) lo the measured sample. On an individual basis, the hypothesized sequence with the closest match to the measured sample (the best metric) is the most likely to have been transmitted. However, over many samples and under the constraint that only certain state transitions are possible, the path tsequence of states) with the minimum cumulative metric has maxihnum likelihood, and this is what the decoder selects.
The system 20 has no a pnon knowledge of the form of the encoder employed in the transmitter.
Performance of the equalizer 21 therefore depends on the accuracy of the estimate of the encoder's state, the charmel impulse response (CIR). FIG. 2 also shows the signals used in estimating the channel impulse response. The objective is to estimate the form of the transversal fnite impulse response filter that would take as its 7 ~8~
input the transmitted inforrnation symbols { a(n) ), and produce at its output the samples taken from the matched filter, { z(n) l . During the transmission of preambles and coded digital verification color codes, the receiver knows the values of {a(n)l. However, at other times, only the estimated values { ad(n) ) are available for use in the channel impulse response estimation process. This dependence leads to a significant perfor-mance-degrading possibility. If decision errors emerge from the equalizer, and these are then used to update the estimate of the channel impulse response, then further deci-sion errors become more probable leading in a circular fashion to fur~her decision errors and breakdown of the equalization process. This phenomena is referred to as a "chan-nel impulse response tracking breakdown". Such difficulties are most Iikely to ar;se at the periods of minimum signal-to-noise ratio, or when the received signal power is at its minimum during reception of a slot.
Within the IS-54 standard, which describes the interface between mobile and base equipment for North Arnerican digital cellular systems, e~ch information time slot is preceded by a known sequence, designated as the "preamble". As viewed by the receiver, therefore, information in the time slot is bounded on both sides by known sequences; the preamble for this slot and the preamble for the subsequent slot. Con-sequently, this equalizer 21 is adapted to rnitigate the effects of a channel impulse re-sponse tracking breakdown. By finding the most probable instant at which the problem might occur, equalizer operation approaches that instant from both forward and a time-reversed directions, both of which begin with l~own inforrnation sequences that are useful for training. Assuming that a channel impulse response tracking breakdownoccurs, this approach minimizes the number of affected symbols by predicting thefailure point and avoiding equalization beyond that point.
At 100 km/hr, which is the maximum speed specified in IS-SS, which describes the mobile unit minimum performance requirements, the average time between fades are on the order of 12 milliseconds. Given time slot durations of about 6.7 milliseconds, there is only a srnall possibility of two significant fades occurring within a time slot.
However, very close to the center of the slot is the code~ digital venfication color code field. Even after a channel impulse response tracking breakdown, the channel impulse response estimator 32 is very likely to recover during processing of the coded digital verification color codes due to the certainty of the transmitted data. Hence, the underly-ing period for which multiple fades are a concern is around 3.5 milliseconds. The chance of more than one deep fade occurring during this time is veIy low. Consequent-3~ ly, tirne-reversed equalization improves bit error rate perfolmance in the digital cellular en~qronment.

8 2 ~
The present equalizer 21 uses a 4-state architecture, corresponding to N = 27 where N is the length of the estimated channel impulse response. This choice assumes that the energy in two (syrnbol-spaced) sarnples of the channel's impulse response dominates. To avoid channel impulse response tracking breakdown problerns, reverse equalization is used for those symbols fo}lowing the minimum power point in a recei~ved time slot.
More specifically, FIG. 3 shows the processing performed in the maximum likelihood sequence estimation based equalizer 21 of FIG. 2. The first step involves finding the location of the power fade (box 51) in terms of symbol number. Processing starts in the forward direction toward the location of the power fade. The symbol num-ber is set to 0 (box 52), and then incremented (box 53). A decision is made whether the symbol then processed is a training symbol (box 54). If the symbol encountered is a training symbol, then training data is inserted (box 57). If a training symbol is not processed, then the equalization trellis is employed to generate metrics and, if possible, a decision (box 55). This is accomplished using equations outlined below. Then it is deterrnined if a decision has been made (box 56). If a decision has been made, then an estimate of the channel impulse response is generated (box 58). If the decision is not made, or once the channel impulse response estimate has been generated, then the sym-bol number is compared to the location of the power fade plus a predetermined number of additional syrnbols (box 59). Processing is then repeated by incrementing the sym-bol number (box 53) and repeating steps (boxes 54-59) until the fade location plus a predeterrnined number of additional symbols has been reached.
Once the desired symbol location is reached in (box 59), then processing is per~formed in the reverse direction star~ing with the prearnble of the next succeeding time slot, namely symbol number 177, for example. The symbol number is set to 178 (box 62), and then decremented (box 63). A decision is made whether the symbol then pro-cessed is a training symbol (box 64). If the symbol encountered is a training symbol, then training data is inserted (box 67). If a training symbol is not processed, then the equalization trellis is employed to generate branch met~ics and a decision box 65).
This is accomplished using the equations outlined below. Then it is determined if a decision has been made (box 66). If a decision has been made, then an estimate of the channei impulse response is generated (box 68). If the decision is no~ made, or once the channel impulse response estimate has been generated, then the symbol number is compared to the location of the power fade less a predetermined n~mber of addihonal 3~ symbols (box 69). Processing is then repeated by decrementing the symbol number (box 63) and repeating steps (boxes 6469) until the fade location less a predetermined number of addi~ional symbols has been reached.

9 2~9~
More particularly, and in operation, samples entering the equalizer 21 may be identified as ztn), and the output decisions may be identified as a(n). The probability of correctness of a(n) depends on location within the bursts. When a(n) is known with certainty the values of a(n), denoted al(n), are used by the channel impulse response S estimator 32 for training. At other times,-the best estimate of a(n) is the output of the traceback decision process of the equalization trellis 31, denoted ad(n).
The equalization trellis 31 operates as follows. Equalization proceeds in the forward direction from the beginning of the preamble up until M symbols after the rninimum power symbol. In the reverse direction, the same occurs with processingcontinuing M symbols beyond the minimum power point. This overlap ensures that trace-back through the trellis in all likelihood converges to a single path by the mini-mum power point.
Traceback for actual decisions does not occur until the completion of the equal-ization process. In addition to final traceback, however, there is a need for tentative decisions during equalization, to provide data estimates for the channel impulseresponse estimation to remain current. A trade-off in determining these tentative deci-sions arises (a) because the more up-to-date the information is, the more up-to-date the channel impulse response estimate can be (remembering that the channel is far fi~om stationary at high speeds), and (b) the higher the number of symbols that are considered before tentative decisions are made, the more accurate the decisions will be; and hence, the lower the probability that errors are introduced into the channel impulse response estimation. In the case of 4-state equ~lization there is very little sensitivity to the num-ber of constraint lengths of delay introduced.
Branch metrics are calculated in the equalizer 21 using the following equation:
br metric [z(k),app_state(l)] = ¦ z(k) - ~ Cn(k) x ah(l,n) n=O
where app sta~e(l) represents a hy~othetical state in combination with potential input data; ah(l,n) is a corresponding transmitted signal (constellation point), C represents the cuFent estimate of the channel's impulse response, and z is the measured ou~ut of the matched filter 27.
Ille channel estima~or 32 utilizes a second ordçr least mean square algorithm todetermine the coefficients of the transversal f;lter 27 that is an estimate of the channel.
Cso~c+l) = cso(k) + K~[z(k) - ~ Cn(k) x â(k-n)] â (k) n=O
Csl(x+l) = Csl(k) + R-Ez(k) - ~, Cn(k) x â(k-n~] â (k-1) n=O

lo ~ 9 ~` ~

Co(lc+l) = Co(k) + cso(k+l) + K2[z(k) - ~, C"(k) x â(k-n)]â (k) n-O
Cl(k+l)= Cl(1c)+Csl(k+l)+K2lz(k) - ~ (k) x â(k-n)]â (k-1), ~I=o where Co(k) and Cl(k) are complex values of estimated channel irnpulse response taps, CSo(k) and Cn(k) are complex intermediate values related to the estimated channel impulse response taps, perrnitting second order operation, Kl and K2 are the real gain values controlling the tracking rate of the channel impulse response estimation process, z(k) are complex symbol spaced sarnpled outputs of the receiver matched filter, and aO are complex estimated or known values of transmitted symbols.
The values Kl and K2 within these equations control the rate of adaptation, and (conversely) the sensitivity to noise and decision errors. Consequently, to minimize the eIror rate, a trade-off between ability to track changes in the channel and degradation in performance due to imperfect input inforrnation is needed to optimiæ the values of Kl and K2. The optimal values of Kl and K2 vary as a function of instantaneous signal to noise ratios, and thus as a function of depth of fade. Therefore, algorithms for modify-ing the values during each burst have been evaluated, with considerable improvement in performance relative to that achievable with constant settings.
One approach for modifying Kl and K2 has provided good perforrnance and is as follows:
1. Set the values of Kl and K2 that will apply at the syrnbol deterrnined to correspond to the deepest fade; Kl_fade.
2. Adjust each value linearly (with preset slope - Kl_slope and K2_slope) to reach the selected values at the fade location, using:
before forward processing - initialize Kl = Kl_fade - Kl_slope fade_location K2 = K2_fade - K2_slope fade_location before reverse processing - initialize Kl = Kl_fade - Kl_slope (17,7 - fade_l~cation) K2 = K2_fade - K2_slope (177 - fade_location) during processing - as each symbol is processed Kl = K1 + Kl_slope K2 = K2 + K2_slope, where Kl_fade is the real value of Kl at the symbol with the maxirnum estimated ~ade dep~t K2_fade is the real value of K2 at Ihe symbol with the maximum estimated fade depth, Kl_slope is the real increment in Kl applied during processing of each symbol, 35 K2_slope is the real increment in K2 applied during processing of each symbol, and , : :

. .

11 2~
fade_location is the syrnbol number at the maximum estimated fade depth, and last_location is the symbol number of the final symbol.
Estimation of the location of the power fade entails use of the received symbolsfrom the matched filter 27, and the settings on the AGC circuit 35 that were active dur-ing reception of those symbols. As the response of the amplifier 22 to the AGC circuit settings is effectively instantaneous, the primary delays in utilizing this inforrnation arise in the matched filter 27. This filter 27 is a linear phase filter (constant delay), so the available input information can be easily transformed into an accurate estimate of the envelope power. This envelope is averaged by a rectangular FIR filter over about ten (10) symbol times, with very good perforrnance.
After completion of acquisition, the c~rrier frequency offset should be less than 200 Hz. To operate without impairrnent, this offset should be on the order of 20 Hz or less. Thus, estimaion of and correction for carrier offset must continue after acquisi-tion. The metnod employed utilizes the fact that when frequency offset occurs, the taps lS of the channel impulse response will rotate consistently at a rate proportional to the off-set. Changes in tap phases over fixed periods, therefore, provide an observable charac-teristic to apply to frequency control. Note that random phase changes occur in addi-tion to these consistent rates of change, so filtering is used to extract the frequency offset. In practice, offsets of around 1000 Hz can be resolved although the maximum expected offset after acquisition is 200 Hz The approach used is as follows:
1. DuIing the reception of each burst, the half of that burst that does not include the deepest fade is selected for tracking. This seheme is aimed at avoidance of the very high rates of change in phase that typically accompany transitions through low signal amplitudes.
2. Two samples of each of the two estimated channel impulse response taps are recorded: just after the preamble (or leading into the postamble if the fade occur~ed during the first half of the slot), and 20 symbols later (or 20 symbols earlier). At a symbol rate of 24300 symbols per second, a 100 Hz offset would result in an average rotation of 29.6 degrees during the 20 symbol period. ~or any rotation in excess of 180 degrees, the observed rotation would be less than 180 degrees but in the opposite direction. This aliasing could impact performance for frequency offsets above about 300 Hz. In typical operation, however, the detriment to performance resulting from such aliasing has proved minimal, due to the anti-aliasing filtering inherent in the tracking. The selection of a sampling window of 20 syrnbols was based on concernabout this aliasing. Otherwise, a longer window would improve noise irnmunity.

12 ~3 3. From information determined during the bit timing fine tuning, the dominant tap is selected. Using the recorded settings for this tap, a phase change is calculated, yielding an estimate of the frequency offset.

4. These estirnates are then filtered over many bursts to reduce the "noise" that 5 anses primarily due to the random (zero mean) presence of Doppler offsets and Gaus-sian noise. The filter output provides an estimate of the carrier offset and can be used to directly update the frequency control hardware. The offset is given by:
f_offset_estimatek+l = (l-KçO)f_offset_estimatek + KfOfreq_observed, where freq_observed is derived from the observed phase change, the constant Kfo con-10 trols the convergence rate of the estimation process, f_offset_estima~ek is the estimatedfrequency offse~ at frame "k", and Kfo is a constant controlling the convergence ràte of the frequency tracking. If f_offset_estimate reaches half the resolution of the frequency source, then a step in frequency is applied,`e.g., if the resolution is 20 Hz and f_offset_estimate exceeds 10 Hz, then a 20 Hz change in reference is applied. At the 1~ same time f_offset_estimate is reinitialized.
Referring to FIG. 5, it illustrates a flow diagram showing the processing per-formed by the equalizer 20 to implement carrier frequency offset compensation. Utiliz-ing an already located fade, a decision (box 100) is made as to whether to use the first or second half of the received slot for frequency of.set estimation. Based on this deci-20 sion, samples are taken twenty symbols apart in the appropriate half of the slot (boxes101, 102). For the selected case, individual taps are compared and the larger is chosen (decisions 103, 104~. The phases of the chosen tap at the selected two times are then subtracted ~boxes 105-108) to produce "freq_observed", a noisy estimate of the offset.
This is filtered (box 109) to generate an accurate estimate of the offset. If an adjustrnent 25 in setting of the frequency control would reduce this offset, then a decision is made to do so (decision 110); and the decision is then implemented (box 111).
The equalizer is reasonably insensitive to errors in bit timing. However, for the following reasons~ symbol tirning adjustments continue during equalizer operation.
The initial estimate produced by acquisition may diffe~. sufficiently from optirnal tirmng 30 so that performance would benefit from adjustrnent. The transmit and receive symbol timing clocks may differ by about 5 ppm, resulting in drift of about ~ S per fra;ne (or a syrnbol every 8 seconds). This drift must be compensated for. In practice, indi-vidual independently-delayed signal paths will randomly rise and diminish in average strength, resulting in situations that would be best catered for by different symbol ~rn-35 ing. Op~imal syrnbol tirning depends on an ability to track these changing situations.
The operation of the symbol timing control is as follows. The approach hassimilarities to the èarly-late gating schemes frequently employed in dirPct-sequence ..
~ , spread spectrum receivers. As each burst is rece~ved, a measure of the error between the expected preamble and the actual received preamble is generated. In addition, in al-ternating frames, similar measures are made on time advanced and retarded versions of the same input samples. If no timing adjustment is necessary, the error generated with S the existing timing should be less (on average) than either of the others. Adjustments are made when this is not the case or there is a consistent dispariy between the ad-vanced and retarded error estimates. This process is simply a search for bit timing that rninimizes the error statistic, as illustrated in FIG. 4. The control loop used includes an estimator of any consistent change in tirning, corresponding to drift with respect to 10 the transmitter. Drift in the order of 10 ppm can be compensated for by this loop.
This search for a minimum may be hampered by the possible presence of a local (non-global) rninimum. In fact, for this statistic the presence of two minima is cornmon (corresponding to the two taps implicit in the equalizer structure - see FIG. 1). The approach taken to resolve this conflict is as follows. The more advanced minimum is 15 presumed to be the preferred sampling time. Multiple minima typically arise when there is a small level of delay spread, i.e., less than about 10 IlS. Under such conditions the ratio of magnitudes of the estimated paths in the (symbol-spaced) channel impulse response differs significantly in the region of the more advanced minimum from that in the more retarded case. Thus, the ratio of tap magnitudes provides a statistic from 20 which to conclude the appropriateness of a selected minimum.
With reference to FIGS. 6A and 6B they show flow diagrarns illustrating the processing performed by the equalizer 20 to implement bit timing control. Inputs (box 80) include the on-time and ~ime-offset samples (z (n) and z offset(n)), and a flag to indicate the direction of the time offset. The on-time sarnples are fed into the equalizer 25 20 just as they are during normal training 83. Similarly, the time offset samples are fed to the equalizer 20 (box 84). In both cases, the branch metrics (on the known correct paths) are accumulated over the latter symbols to provide measures (ERRORC,~ andERROR OF~SETCUm) of the degree to which the samples match expectations.
In a separate process the magnitudes of each of h~O taps estimated as the chan-30 nel impulse response at the end of the tr~ining process are calculated (box 85). Averag-ing the ratio of these taps over a number of frames (boxes 86-89) perrnits a judgement to be made as to whether the bit timing has selected an inappropriate local minimulIL If a threshold (box 90) is reached, then bit timing will be advanced by a full symbol tirne (box 91). Taking account of the relative time at which samples were taken (box 92~, 35 the ERRORCUm and ERROR OFFSET~ "m measures are combined to generate a noisy estimate of an appropriate timing adjustment (boxes 93, 94). This estimate is then 14 206~9g~
filtered (box 95) to generate an actual timing offset adjustment. To compensate for consistent drift, an additional terrn "drift_est" monitors and compensates for this effect.
Thus there has been described a maximum likelihood sequence estimation based equalization method for use in mobile digital cellular receivers. It is to be understood 5 that the above-described embodiments are merely illustrative of some of the many spe-cific embodiments which represent applications of the principles of the present inven-tion. Clearly, numerous and other arrangements can be readily devised by those skil}ed in the art without departing from the scope of the invention.

~' ~' ~ " ` ' '

Claims

1. In a digital communications receiver, a method of processing samples received from a delay-spread, fading channel, which samples are associated with a block of data, including a sequence of symbols, transmitted within a time slot, which method is adapted to make data decisions, said method comprising the steps of:
storing samples received during a time slot in a time sequential manner;
estimating the location within the time slot at which decision errors are most probable which is the location of maximum fade depth;
processing the stored samples, starting with the first received sample and proceeding in a forward direction with respect to the time sequence in which thesamples were stored, beyond the estimated location of the maximum fade depth, using a preselected maximum likelihood sequence estimation procedure to generate estimates of the transmitted data;
processing the stored samples, starting with the final received sample and proceeding in a reverse direction with respect to the time sequence in which the samples were stored, beyond the estimated location of maximum fade depth, using the maximum likelihood sequence estimation procedure to generate estimates of the transmitted data;
simultaneous with the above two processing steps, generating estimates of the transmission channel impulse response, which are used in the maximum likelihood sequence estimation procedure; and processing the output of the preceding estimate generating steps to generate data decisions.

2. The method of Claim 1 wherein the steps of generating the channel impulse response estimates comprises using variable tap coefficients that are determined by:
estimating tap settings for the estimated channel impulse response by minimizing the square of the difference between actual received samples and those synthesized by passing known transmitted signals through the estimated channel, and wherein the processing is done in an iterative manner by combining previous estimates of channel impulse response and new estimates thereof based on recent estimates, and by varying the ratio of the contributions from the previous and new estimates as a function of location within the time slot.

3. The method of Claim 2 wherein the the values of coefficients K1 and K2 are varied as a function of symbol location (k) by solving the following equations for (k+1) and Cl(k+1):

where C0(k) and C1(k) are complex values of estimated channel impulse response taps, CSO(k) and Cn(k) are complex intermediate values related to the estimated channel impulse response taps, permitting second order operation, K1 and K2 are the real gain values controlling the tracking rate of the channel impulse response estimation process, z(k) are complex symbol spaced sampled outputs of the receiver matched filter, and a(k) are complex estimated or known values of transmitted symbols.

4. The method of Claim 3 wherein the values of K1 and K2 comprising the variable coefficients are modified by:
setting the values of K1 and K2 that will apply at the symbol determined to correspond to the estimated location of a maximum fade depth (K1_fade and K2_fade, respectively); and adjusting each value linearly (with preset slope - K1_slope and K2_slope) to reach the selected values at the location of the maximum fade depth, in accordance with the following equations:
before forward processing - initialize K1 = K1_fade - K1_slope ? fade_location K2 = K2_fade - K2_slope fade_location before reverse processing - initialize Kl = K1_fade - K1_slope ? (last_location - fade_location) K2 = K2_fade - K2_slope (last_location - fade_location) during processing - as each symbol is processed K1 = K1 + K1_slope K2 = K2 + K2_slope where K1_fade is the real value of K1 at the symbol with the maximum estimated fade depth, K2_fade is the real value of K2 at the symbol with the maximum estimatedfade depth, K1_slope is the real increment in K1 applied during processing of each symbol, K2_slope is the real increment in K2 applied during processing of each symbol, and fade_location is the symbol number at the maximum estimated fade depth, and last_location is the symbol number of the final symbol.

5. The method of Claim 2 which further comprises a symbol timing adjustment method comprising the steps of:
using a subset of the samples received during the time slot, and generating therefrom an error measurement comprising a prime measure of the degree to which the estimated channel impulse response matches the actual channel impulse response;
generating a plurality of similar measures utilizing simultaneously recorded samples having different time offsets wherein at least one sample is advanced and one sample is retarded in time relative to the prime measure; and determining, by filtering a plurality of such measures, a bit time setting that minimizes the above error measurement and adjusting the sampling to reflect the newly determined bit time setting.

6. The method of Claim 2 which further comprises a carrier offset tracking method comprising the steps of:
recording at least two samples of at least one tap within the estimated channel impulse response at selected symbol locations within the time slot;
generating frequency offset estimates at each of a plurality of time slots by observing and determining the phase difference in each time slot between the at least two samples;
combining this plurality of frequency offset estimates by using a filtering process to generate a precise frequency offset estimate; and adjusting a controllable frequency source to compensate for the precise frequency offset estimate.

7. The method of Claim 2 which further comprises a carrier offset tracking method comprising the steps of:
during the reception of each time slot, a first portion of the time slot that does not include the estimated location of maximum fade depth;
recording two samples of each of the two estimated channel impulse response taps, a first just after the preamble or in the postamble, depending upon the location of the estimated location of maximum fade depth, and a predetermined number of symbols later or earlier, depending upon the location of the estimated location of maximum fade depth;
from information determined during the bit timing, selecting a dominant tap;
using recorded settings for the selected dominant tap, calculating a phase change, thereby yielding an estimate of the frequency offset;
filtering the estimate over many bursts and generating an estimate of the carrier offset and adapting the generated estimate of carrier offset as a feedback signal, thereby adjusting carrier tracking.

8. The method of Claim 7 wherein the offset is given by:
f_offset_estimatek+1= (1-Kfo)f_offset_estimatek + Kfofreq_observed, where freq_observed is derived from the observed phase change, the constant Kfo controls the convergence rate of the estimation process, f_offset_estimatek is the estimated frequency offset at frame "k", and Kfo is a constant controlling the convergence rate of the frequency tracking.

9. The method of Claim 8 which further comprises a carrier frequency offset compensation method comprising the steps of:
utilizing an already located fade, making a decision as to whether to use the first or a second portion of the received time slot for frequency offset estimation;
based on this decision, taking samples a predetermined number of symbols apart in the appropriate portion of the time slot;
for the selected case, comparing individual taps and choosing the larger one;
subtracting the phases of the chosen tap at the selected two times to produce "freq_observed", a noisy estimate of the frequency offset;
filtering the freq_observed estimate to generate an accurate estimate of the frequency offset; and adjusting the setting of the frequency control if this would reduce the frequency offset.

10. The method of Claim 5 wherein the block of data further includes a preamble and a postamble and which further comprises a carrier offset tracking method comprising the steps of:
during the reception of each time slot, selecting the half of the time slot that does not include the estimated location of maximum fade depth;
recording two samples of each of the two estimated channel impulse response taps, a first just after the preamble or in the postamble, depending upon the location of the estimated location of maximum fade depth, and a second predetermined number of symbols later or earlier, depending upon the location of the estimated location of maximum fade depth;
from information determined during the bit timing, selecting a dominant tap;
using recorded settings for the selected dominant tap, calculating a phase change, yielding an estimate of the frequency offset;
filtering the estimate of the frequency offset over many bursts to generate an estimate of the carrier offset adapted for use as a feedback signal to adjust carrier tracking.

11. In a digital cellular receiver, a method of processing samples received froma delay-spread, fading channel, that are associated with a block of data that was transmitted within a time slot, said method comprising the steps of:
processing the samples starting with a first received sample and proceeding beyond a final received sample of the time slot using a preselected maximum likelihood sequence estimation process to generate estimates of the transmitted data;
simultaneously producing estimates of the characteristics of the impulse response of the channel which are used in the maximum likelihood sequence estimation process; and using the results of the preceding two processing steps to generate final decisions as to the transmitted data.

12. In a digital cellular receiver, a method of processing samples received from a delay-spread, fading channel, that are associated with a block of data that was transmitted within a time slot, said method comprising the steps of:
processing the samples starting with a first received sample and proceeding beyond a final received sample of the time slot using a preselected maximum likelihood sequence estimation process to generate estimates of the transmitted data;
simultaneously producing estimates of the characteristics of the impulse response of the channel which are used in the maximum likelihood sequence estimation process; and using the results of the preceding two processing steps to generate final decisions as to the transmitted data.

13. In a digital communications receiver, a method of compensating for fading in sequential samples transmitted within a time slot, the method comprising the steps of storing samples received during a time slot in a time sequential manner;
estimating the location within the time slot of maximum fade depth;
processing the stored samples, starting with the first received sample and proceeding in a forward direction with respect to the time sequence in which the samples were stored, beyond the estimated location of the maximum fade depth, using a preselected maximum likelihood sequence estimation procedure to generate estimates of the transmitted data;
processing the stored samples, starting with the final received sample and proceeding in a reverse direction with respect to the time sequence in which the samples were stored, beyond the estimated location of maximum fade depth, using the maximum likelihood sequence estimation procedure to generate estimates of the transmitted data; and processing the output of the preceding estimate generating steps to compensate for fading of the samples.

14. The method of Claim 13 comprising generating transmission channel impulse response estimates for use in the maximum likelihood sequence estimation procedures.

15. The method of Claim 14 wherein the steps of generating the channel impulse response estimates comprise using variable tap coefficients that are determined by estimating tap settings for the estimated channel impulse response by minimizing the square of the difference between actual received samples and those synthesized by passing known transmitted signals through the estimated channel, and wherein the processing is done in an iterative manner by combining previous estimates of channel impulse response and new estimates thereof based on recent estimates, and by varying the ratio of the contributions from the previous and new estimates as a function of location within the time slot.

16. The method of Claim 14 which further comprises the steps of:
using a subset of the samples received during the time slot, and generating therefrom an error measurement comprising a first measure of the degree to which the estimated channel impulse response matches the actual channel impulse response;
generating a plurality of similar measures utilizing simultaneously recorded samples having different time offsets wherein at least one sample is advanced and one sample is retarded in time relative to the first measure;
and determining, by filtering a plurality of such measures, a bit time setting that minimizes the above error measurement and adjusting the sampling to reflect the newly determined bit time setting.

17. The method of Claim 14 which further comprises the steps of:
recording at least two samples of at least one tap within the estimated channel impulse response at selected locations within the time slot;

generating frequency offset estimates at each of a plurality of time slots from the phase difference in each time slot between the at least two samples;
combining this plurality of frequency offset estimates by, for example, using a filtering process to generate a precise frequency offset estimate; and adjusting a controllable frequency source to compensate for the precise frequency offset estimate.

18. The method of Claim 14 wherein the samples further include a preamble and a postamble and which further comprises the steps of:
during the reception of each time slot, selecting the half of the time slot that does not include the estimated location of maximum fade depth;
recording two samples of each of the two estimated channel impulse response taps, a first just after the preamble or in the postamble, depending upon the location of the estimated location of maximum fade depth, and a second predetermined number of symbols later or earlier, depending upon the location of the estimated location of maximum fade depth;
from information determined during the bit timing, selecting a dominant tap;
using recorded settings for the selected dominant tap, calculating a phase change, yielding an estimate of the frequency offset;
filtering the estimate of the frequency offset over a plurality of bursts to generate an estimate of the carrier offset adapted for use as a feedback signal to adjust carrier tracking.

19. In a digital communications receiver in which samples are received sequentially within a time slot of a transmission channel, a method for adjusting sample timing comprising the steps of:
storing samples received during a time slot in a time sequential manner;

generating estimates of the transmission channel, impulse response;
using a subset of the samples received during the time slot, and generating therefrom an error measurement comprising a first measure of the degree to which the estimated channel impulse response matches the actual channel impulse response;
generating a plurality of similar measures utilizing simultaneously recorded samples having different time offsets wherein at least one sample is advanced and one sample is retarded in time relative to the first measure;
and determining, by filtering a plurality of such measures, a bit time setting that minimizes the above error measurement and adjusting the sampling to reflect the newly determined bit time setting.

20. In a digital communications receiver in which samples are received sequentially within a time slot of a transmission channel, a carrier offset tracking method comprising the steps of:
storing samples received during a time slot in a time sequential manner;
generating estimates of the transmission channel, impulse response;
recording at least two samples of at least one tap within the estimated channel impulse response at selected symbol locations within the time slot;
generating frequency offset estimates at each of a plurality of time slots from the phase difference in each time slot between the at least two samples;
combining this plurality of frequency offset estimates by, for example, using a filtering process to generate a precise frequency offset estimate; and adjusting a controllable frequency source to compensate for the precise frequency offset estimate.

21. In a digital communications receiver in which samples, which include preambles and postambles are received sequentially within a time slot of a transmission channel, a carrier offset tracking method comprising the steps of:
storing samples received during a time slot in a time sequential manner;
generating estimates of the transmission channel, impulse response;
during the reception of each time slot, selecting the half of the time slot that does not include the estimated location of maximum fade depth;
recording two samples of each of the two estimated channel impulse response taps, a first just after the preamble or in the postamble, depending upon the location of the estimated location of maximum fade depth, and a second predetermined number of symbols later or earlier, depending upon the location of the estimated location of maximum fade depth;
from information determined during the bit timing, selecting a dominant tap;
using recorded settings for the selected dominant tap, calculating a phase change, yielding an estimate of the frequency offset;
filtering the estimate of the frequency offset over many bursts to generate an estimate of the carrier offset adapted for use as a feedback signal to adjust carrier tracking.