Synchronisation in a multicarrier receiver with guard interval carrelation

This application, entitled "IMPROVED TIME TRACKING ALGORITHM VIA GUARD INTERVAL CORRELATION FOR OFDM SIGNALS", was filed on November 15, 2005 and transferred to this transferee. Claims priority over US patent application provisional number 60 / 737,087 incorporated herein by reference.

The present disclosure relates generally to communications, and in particular to techniques for time tracking in receivers in communication systems.

Orthogonal Frequency Division Multiplexing (OFDM) is a multi-carrier modulation technique that provides good characteristics for some wireless environments. OFDM divides the entire system bandwidth into a large number (K) orthogonal frequency subbands, which are also called subcarriers, tones, bins, etc. With OFDM, each subband is associated with its own subcarrier that is modulated by the data. Up to K modulation symbols are sent over the K subband during each OFDM symbol period.

In OFDM systems, transmitters typically transform the modulated symbols of each OFDM symbol period into the time domain by K-point inverse fast Fourier transform (IFFT) or inverse discrete Fourier transform (IDFT) to obtain a K time domain chip. .. To reduce delay spread in the communication channel, the transmitter repeats several K chips to form the OFDM symbol. The repeated part is commonly referred to as the guard interval or cyclic prefix. The guard section is used to reduce intersymbol interference (ISI) and intercarrier interference (ICI) caused by delay spread, which is between the earliest and slowest arrival signal paths in the receiver. It is a time difference.

The receiver performs complementary processing and removes the guard interval in each received OFDM symbol. The receiver then transforms the K-time domain sample of each received OFDM symbol into the frequency domain by the K-point Fast Fourier Transform (FFT) or Discrete Fourier Transform (DFT) to obtain the K-received symbols of the K-subband. The receiver then makes a detection on the received symbol to restore the transmitted modulation symbol. The receiver generally maintains a time tracking loop that determines the proper FFT window placement for each received OFDM symbol. This FFT window indicates which specimens should be maintained and which specimens should be discarded. The data detection characteristics are greatly affected by the arrangement of the FFT windows.

Therefore, there is a need in the art for time tracking techniques in the receiver to achieve FFT window placement.

Outline of the invention

Techniques for performing time tracking in the receiver are described here. Channel impulse response estimates for communication channels are obtained, for example, based on receiving pilots. The first arriving path (FAP) and the last arriving path (LAP) are detected based on channel impulse response estimates. FAPs and LAPs are detected correctly, and the detected FAPs and LAPs are true FAPs and LAPs, respectively. However, if the delay spread of the communication channel is excessive, the FAPs and LAPs will not be detected accurately, and the detected FAPs and LAPs will be exchanged and correspond to the true LAPs and FAPs, respectively.

Two assumptions are evaluated to resolve the ambiguity in the detected FAPs and LAPs. The first assumption corresponds to the correctly detected FAPs and LAPs, and the second assumption corresponds to the incorrectly detected FAPs and LAPs. For each assumption, the hypothetical FAP and LAP are determined based on the detected FAP and LAP, and the correlation window is determined based on the hypothetical FAP and LAP. The correlation window for each assumption covers all or part of the guarded section of the hypothetical FAP and / or all or part of the guarded section of the hypothetical LAP. For each assumption, the correlation is between the first segment of the received data and the second segment of the received data in the correlation window, which are K-samples apart (K is the period of the useful part of the OFDM symbol). .. The correct assumption is determined based on the correlation results for the two assumptions. The receiver timing is updated based on the assumptions FAP and LAP for the correct assumptions. OFDM demodulation is based on updated receiver timing, for example, the FFT window is placed accurately based on receiver timing.

Various embodiments and examples of the present invention are described in further detail below.

The features and properties of the present invention will become further apparent from the detailed description beginning below taken in connection with the drawings in which similar reference numerals are correspondingly identical throughout.

Detailed explanation

The term "exemplary" is used herein to mean "useful as an example, instance, or illustration." The embodiment described herein as "typical" is not necessarily construed as favorable or advantageous over other embodiments.

The time tracking techniques described herein are used in a variety of communication systems such as OFDM systems, orthogonal frequency division multiple access (OFDMA) systems, single carrier frequency division multiple access (SC-FDMA) systems, and so on. The OFDM system utilizes OFDM. SC-FDMA systems are interleaved FDMA (IFDMA), which is transmitted over subbands distributed over the system bandwidth, localized FDMA (LFDMA), which is transmitted over one block of adjacent subbands, or adjacent subbands. Utilizing Extended FDMA (EFDMA) transmitted over a large number of blocks Generally, modulated symbols are sent by OFDM in the frequency region and by SC-FDMA in the time region. The SC-FDMA symbol contains a guard interval generated in the same way as the guard interval of the OFDM symbol. For clarity, time tracking technology specifically describes OFDM-based systems below.

FIG. 1 shows a block diagram of transmitter 110 and receiver 150 in an OFDM system. At transmitter 110, the transmit (TX) data processor 120 processes (eg, formats, encodes, interleaves, and symbol maps) traffic data and produces data symbols. As used here, the data symbol is a modulation symbol for traffic data, and the pilot symbol is a modulation symbol for the pilot, which is a priori known data by both the transmitter and receiver. Yes, and the zero symbol is the signal value of zero.

The OFDM modulator 130 receives data symbols and pilot symbols, respectively, and multiplexes them on the data and pilot subbands, performs OFDM modulation as described below, and provides OFDM symbols during each OFDM symbol period. The OFDM symbol period is the duration of one OFDM symbol, and is also referred to as the symbol period. Transmitter unit (TMTR) 132 receives and processes OFDM symbols (eg, analog conversion, amplification, filtering, and high frequency conversion) and produces a modulated signal, which is transmitted by antenna 134 to receiver 150. ..

At receiver 150, antenna 152 receives the modulated signal from transmitter 110 and provides the received signal to receiver unit (RCVR) 154. The receiver unit 154 tunes the received signal (eg, filters, amplifies, lower frequency conversions, and digitizes) and provides a received sample. As described below, the OFDM demodulator (Demod) 160 processes a received sample and obtains a K-received symbol for the K-subband during each OFDM symbol period. Received symbols include received data symbols in the data subband and received pilot symbols in the pilot subband. The OFDM demodulator 160 performs time tracking to properly position the FFT window during each OFDM symbol period. The OFDM demodulator 160 also demodulates / detects the received data symbol by channel estimation in order to obtain a data symbol estimate, which is an estimate of the data symbol sent by transmitter 110. The receive (RX) data processor 170 processes data symbol estimates (eg, symbol reverse mapping, inverse interleaving, and decoding) to obtain the decoded data there. In general, the processing by the OFDM demodulator 160 and the RX data processor 170 is complementary to the processing by the OFDM modulator 130 and the TX data processor 120 in the transmitter 110, respectively.

Controllers / processors 140 and 180 control the operation of various processing units at transmitter 110 and receiver 150. The memories 142 and 182 store the data and program code of the transmitter 110 and the receiver 150, respectively.

FIG. 2 shows a block diagram of the OFDM modulator 130 of the transmitter 110 of FIG. Within the OFDM modulator 130, the symbol-to-subband mapping unit 210 receives the data and pilot symbols, maps them to the subbands used for the data and pilots, respectively, and zero symbols to the unused subbands. And provide a K transmission symbol for the entire K subband. Each transmission symbol is a data symbol, pilot symbol, or zero symbol for each OFDM symbol period. During each OFDM symbol period, the unit 212 transforms the K transmission symbol into the time domain by the K point IFFT / IDFT and provides the transform symbol containing the K time domain chip. Each chip is a complex value that should be transmitted in one chip period. The parallel-to-series (P / S) converter 214 serializes the K chip of each conversion symbol. The guard section insertion unit 216 repeats a part (or G chip) of each conversion symbol there to form an OFDM symbol containing a K + G chip. Each OFDM symbol contains a K-chip for useful parts and a G-chip for guard sections. Guard sections are used to reduce ISI and ICI caused by delay spread in the communication channel. The length of the guard interval G determines the maximum delay spread that the receiver can withstand at moderate moving speeds without suffering from ISI and ICI.

FIG. 3 shows a block diagram of an embodiment of the OFDM demodulator 160 of the receiver 150 of FIG. Within the OFDM demodulator 160, the preprocessor 310 processes the sample received from receiver unit 154 and provides an input sample. The preprocessor 310 performs automatic gain control (AGC), timing acquisition, filtering, sampling rate conversion, DC (DC) offset elimination, frequency error estimation and elimination, and / or other functions. Unit 312 removes the guard interval in each received OFDM symbol based on the FFT start pointer and provides a K input sample for that OFDM symbol. The FFT start pointer controls the placement of the FFT window for each received OFDM symbol.

For each received OFDM symbol, unit 314 performs a K point FFT / DFT on the K input sample and provides the K frequency domain received symbol over the K subband. Channel estimator 318 obtains channel estimates based on the received pilot symbol. The channel estimates are time domain channel impulse response estimates and / or frequency domain channel frequency response estimates. The data demodulator 316 demodulates / detects the received data symbol by channel estimation and provides the data symbol estimate.

As described below, the time tracking loop 320 performs time tracking, determines the timing of each received OFDM symbol, and provides an FFT start pointer. Although not shown in FIG. 3 for simplicity, the OFDM demodulator 160 includes processing units for frame detection, frame synchronization, frequency tracking, and other functions.

The channel estimator 318, the time tracking loop 320, and the other units in the OFDM demodulator 160 operate at the sample rate. These units also discard samples to reduce complexity and process at a lower rate. These units also oversample the received signal to achieve better resolution and process at higher speeds. For clarity, the description below assumes processing at that sample rate, and various quantities, constants, and thresholds are given for sample rate processing.

The receiver performs time tracking in various ways and based on various forms of information sent by the transmitter. For example, the transmitter transmits the pilot over the N subband, which is evenly distributed over the K subband (provided that 1 <N <K). The receiver receives the OFDM symbol containing the pilot, removes the guard interval, and performs a K-point FFT / DFT on the useful part of the received OFDM symbol to obtain the N-received pilot symbol for the N-pilot subband. The receiver then unmodulates the receive pilot symbol to obtain the N-channel gain, and then the N-channel tap (which is n = 0, ..., h for N-1)._nPerform an N-point FFT / DFT on the N-channel gain to obtain a channel impulse response estimate containing). The receiver also obtains channel impulse response estimates in other ways known in the art. In the embodiments described below, the receiver performs time tracking based on channel impulse response estimates.

FIG. 4 shows a block diagram of an embodiment of the time tracking loop 320 of FIG. Within the time tracking loop 320, unit 410 receives N-channel taps from channel estimator 318 for channel impulse response estimation, and squares each channel tap. magnitude) is calculated. Unit 410 filters the square of the channel tap over a number of symbol periods. Unit 410 provides a channel power profile containing N (filtered or unfiltered) squared values for N-channel taps. The detector 412 detects the first arrival path (FAP) and the last arrival path (LAP) based on the channel power profile, and provides the detected FAPs and LAPs as described below. Unit 414 resolves ambiguities in the detected FAPs and LAPs and provides the output FAPs and LAPs. Center of Gravity (CM) Detector 416 determines the centroid of the channel power profile based on the output FAP and LAP. The adder (summer 418) subtracts the timing target from the detected centroid and provides the timing error. The timing goal is the target position of the FFT window, and is a programmable value. The timing error represents the error between the detected center of gravity and the target position for the FFT window.

The loop filter 420 filters for timing errors and adjusts the timing. For the embodiment shown in FIG. 4, the loop filter 420 implements a first-order low-pass filter. The multiplier 422 multiplies the timing error by the gain. Adder 424 aggregates the output of multiplier 422 with the output of delay unit 426 and adjusts the timing. Delay unit 426 stores the output of adder 418 for the next loop update. Other designs and / or other transfer functions are also used for the loop filter 420. The limiter 430 limits timing adjustments within a predetermined range and provides an Adv / Ret output that indicates how much the FFT window should be moved. Adder 434 aggregates the progress / delay output along with the current FFT start pointer from delay unit 436, and provides an updated FFT start pointer. Delay unit 436 stores the updated FFT start pointer for the next loop update.

The detector 412 detects FAP and LAP in various ways. For clarity, special techniques for detecting FAPs and LAPs based on channel power profiles are described below. For this technique, the energies of all channel taps in the sliding window are first calculated for different tap positions k as follows:<maths num="1"><img file="JP4955693B2_D0001.tif" /></maths>

(k = 0, ..., N-1) However, E_kIs the energy of the channel tap in the sliding window at the tap position, W is the width of the sliding window, and Mod N stands for modulo-N operation.

The sliding window width is chosen to be less than or equal to or equal to the length of the channel impulse response estimate, or W N / 2. The estimated channel impulse response length is generally less than, equal to, or N G for the guard interval, which is generally chosen so that the guard interval is greater than the channel delay spread to avoid ISI and ICI. Because it is done. Equation (1) essentially moves the sliding window in a circular shape over the channel power profile, and for each tap position k, calculates the energy of the W channel tap in the sliding window. The sliding window wraps around the front of the channel power profile as it reaches the end of the channel power profile.

The finite difference is then calculated based on the tap energy for each tap position as follows:<maths num="2"><img file="JP4955693B2_D0002.tif" /></maths>

(k = 0, ..., N-1) However, Q is a finite difference order, and D_nIs a finite difference value for the tap position.

A 2Q long sliding window is used to calculate a finite difference. This sliding window moves in a circular shape over the channel power profile. Finite difference D for each tap position n_nIs calculated as the difference obtained by subtracting the binding energy of the second half of the sliding window from the binding energy of the first half of the sliding window.

The metric used to detect FAP and LAP is defined as follows based on tap energy and finite difference: S_FAP(k) = α₁ E_k+ (1-α₁) D_{(k-Q + 1) mod N}(k = 0, ···, N-1) Equation (3) S_LAP(k) =-α₂ E_k+ (1-α₂) D_{(kQ) mod N}(k = 0, ···, N-1) Equation (4) However, α₁And α₂Are the coefficients used for FAP and LAP detection, respectively. S_FAP(k) is the metric used for FAP detection at tap position k, and S_LAP(k) is the metric used for LAP detection at tap position k.

Coefficient α₁Determines the sensitivity of FAP detection, and is selected to detect FAP if it is the last large value before the drop in the region of cumulative energy close to maximum. Similarly, the coefficient α₂Determines the sensitivity of LAP detection, and is selected so that LAP is detected if it is the first large value before the increase in the region of cumulative energy close to maximum. Coefficient α₁And α₂Are selected to provide good performance for FAP and LAP detection, respectively, and are determined based on computer simulations, experimental measurements, etc.

The location of FAP is S_FAPBased on (k):<maths num="3"><img file="JP4955693B2_D0003.tif" /></maths>

However, FAPd is an index of detected FAP with sufficient intensity. In equation (5), S_FAPThe maximum value of (k) is identified first, and FAPd is S_FAPIt is set to the index that gives the maximum value of (k).

LAP location is S_LAPBased on (k):<maths num="4"><img file="JP4955693B2_D0004.tif" /></maths>

LAPd = (k_min+ W-1) mod N expression (7) However, LAPd is an index of detected LAP with sufficient intensity. In equations (6) and (7), S_LAPThe minimum value for (k) is identified first and S_FAPThe exponent that gives the minimum value for (k) is k_minExpressed as, and LAPd is k_minIt is set to the W-1 tap position on the right side of. The mod N operation in Eq. (7) constrains LAPd to the range 0 to N-1. In the following description, FAP and LAP represent true FAP and LAP, respectively, and FAPd and LAPd represent detected FAP and LAP, respectively.

Figures 5A-5C show FAP and LAP detection for typical channel impulse response estimation. For this example, N = 16, and as shown in Figure 5A, the channel impulse response estimation involves two large channel taps at tap positions 3 and 8. Figure 5B shows the energy E for the channel impulse response estimates shown in Figure 5A._kGraph 512 and graph 514 of finite difference (according to W = 8 and Q = 2) are shown. Figure 5C shows α₁= α₂Weighing standard S by = 0.5_FAPGraph 516 of (k) and measurement standard S_LAPThe graph 518 of (k) is shown.

As shown in Figure 5B, FAP is energy E_kConfirmed by detecting the fall in graph 512 of, which is also a finite difference D_nIt is confirmed by detecting the peak in. D_nThe finite difference operation of increases noise. Therefore, the finite difference D_nAnd energy E_k Weighted sum is the weighing standard S_FAPUsed as (k). Similarly, LAP is energy E_kConfirmed by detecting the rise in graph 512 of, which is also a finite difference D_nIt is confirmed by detecting the bottom price (trough) in the graph 514 of. Finite difference D_nAnd energy E_k The sum of loads is the metric S to reduce the effect of increased noise._LAPUsed as (k). For the example shown in Figures 5A-5C, S_FAPThe maximum value of (k) occurs at the exponent k = 3, and FAP is detected to be at FAPd = 3. S_LAPThe minimum value of (k) occurs at the exponent k = 1, and LAP is detected to be LAPd = 1 + 8-1 = 8.

Figures 5D-5F show FAP and LAP detection for another typical channel impulse response estimate involving two large channel taps at tap positions 3 and 13. Figure 5E shows energy E_kGraph 522 and finite difference D_n Graph 524 is shown. Figure 5F shows the measurement standard S_FAPGraph 526 of (k) and measurement standard Measurement standard S_LAPThe graph 528 of (k) is shown. For the example shown in Figures 5D-5F, S_FAPThe maximum value of (k) occurs at the index k = 13, and FAP is detected to be at FAPd = 13. S_LAPThe minimum value of (k) is generated in the index k = 12, LAP is LAPd = (12 + 8-1) mod16 = 3 near Rukoto is detected.

For clarity, special techniques for detecting FAP have been described above. This technique provides incorrect FAP and / or LAP under certain conditions. For example, if the sliding window length is smaller than the channel delay spread, it is possible to select the intermediate channel path as a possible candidate for FAP or LAP. In addition, depending on the relative position of the channel paths and / or their relative power, this detection method with W = N / 2 detects different candidates for FAP and LAP. False detection of FAP and LAP is, for example, a point D where the front center of gravity is predetermined._midIt can be avoided by assuming that it was properly located around. Therefore, for the current channel estimation, which signal path is D_midWhich signal path will appear later, and which signal path is D_midThe first decision is made as to whether it comes before, but there is a lie in the early channel estimation. Assuming that the channel content is likely to appear early and equally late, the disconnect between the past and the future is D._minIt is assumed to be set to + N / 2, and any channel taps after this cut point arrive early, which is false. The channel content is just D_mid+ max {Δ_hNote that it existed before to} / 2 (here Δ_hIs a channel delay spread). D_mid+ max {Δ_h} / 2 and D_midThe area between + N / 2 indicates the search area for slow channel content. In general, FAP and LAP detection is performed in different ways and for different channel conditions.

Channel delay spread Δ_hIs less than or equal to the guard interval, or Δ_hWhen G / 2, the FAP and LAP detection methods described above in equations (1) to (7) provide relatively accurate FAP and LAP detection. However, when the delay spread is longer than half of the guard interval, the decisions are exchanged so that LAPd = FAP and FAPd = LAP. Therefore, the detected FAP may or may not be a true FAP, and the detected LAP may or may not be a true LAP. The accuracy of FAPd and LAPd determination depends on the actual channel tap, channel delay spread, and possibly other factors. Therefore, it is desirable to realize accurate detection of FAP and LAP.

FAP and LAP are also detected by other methods. Regardless of the method used for FAP and LAP detection, there is ambiguity as to whether the detected FAPs and LAPs are correct or exchanged.

Ambiguity in FAP and LAP detection is resolved by taking advantage of the relationship between the start of the FFT window, which is indicated by the FFT start pointer, and the resulting channel delay profile. In particular, the location of FAPs and LAPs in channel impulse response estimation depends on (1) the propagation delay of FAPs and LAPs, and (2) the placement of FFT windows. This relationship is illustrated in the following two examples.

Figure 6A shows the typical transmission of an OFDM symbol over a communication channel with multiple signal paths. In general, each signal path has some complex gain and some propagation delay, both of which are determined by the channel environment. For simplicity, the gains of FAP and LAP are considered equal in this example. Communication channel delay spread is Δ_hIs the difference between the propagation delays of FAP and LAP.

The OFDM symbol contains useful parts and guard sections. The receiver obtains an OFDM symbol copy via each signal path. Each OFDM symbol copy is constantly standardized by the complex gain of the associated signal path, and further delayed by the propagation delay of that signal path. For simplicity, FIG. 6A shows only the first OFDM symbol received via FAP and the last OFDM symbol received via LAP. The start of the last OFDM symbol copy is Δ from the start of the first OFDM symbol copy._hOnly the specimen is delayed. The received OFDM symbol is a superposition of all the OFDM symbol copies in the receiver.

Figure 6A also shows the placement of the FFT window for the OFDM symbol. The start of the FFT window is indicated by the FFT start pointer. The FFT window has the width of the K sample and determines which sample is selected for the next process. For the example shown in Figure 6A, the start of the FFT window is delta sample away from the first sample in the useful portion of the OFDM symbol copy for FAP, and Δ + from the first sample in the useful part of the OFDM symbol copy for LAP. Δ_hSpecimens are separated.

Figure 6A also shows the area without ISI / ICI, which is the overlapping area of the guarded sections of all OFDM symbol copies in the receiver. If the FFT start pointer is in the region without ISI / ICI, it is desirable because the appropriate K sample is selected for processing and does not encounter ISI / ICI. The purpose of the time tracking loop is to keep the FFT start pointer in the region without ISI / ICI.

Figure 6B shows the typical transmission channel impulse response estimation shown in Figure 6A. The channel impulse response of the communication channel includes a channel tap for each signal path. For simplicity, Figure 6B shows Δ and Δ + Δ for FAP and LAP._hShows only two channel taps of the index of. The size of each channel tap is determined by the complex gain of the associated signal path. The location or exponent of each channel tap is determined by the propagation delay of the associated signal path and FFT start pointer.

Figure 6B essentially shows the channel delay profile of a communication channel. For this channel delay profile, the FFT start pointer is mapped to the origin. The channel gain of each signal path is mapped to an exponent determined by the distance between the FFT start pointer and the first sample in the useful part of the OFDM symbol copy of that signal path. The channel delay profile contains information indicating the start of channel impulse response estimation.

Figure 6C shows G-Δ_hIt shows another typical transmission of an OFDM symbol over a communication channel with multiple signal paths with a delay spread of. For the example shown in Figure 6C, the start of the FFT window is Δ + Δ from the first sample in the useful part of the OFDM symbol copy of the FAP._hSpecimens are separated, and Δ + G samples are separated from the first sample in the useful part of the OFDM symbol copy of the LAP.

Figure 6D shows the typical transmission channel impulse response estimation shown in Figure 6C. This channel impulse response estimate corresponds to LAP and FAP Δ and Δ + Δ_hIncludes two channel taps in the index of. In this example, the first sample in the useful part of the OFDM symbol copy of the LAP is greater than or equal to G from the start of the FFT. This results in a LAP channel tap that wraps around the exponent Δ and appears, which is to the left of the FAP channel tap.

As shown in FIGS. 6A-6D, the channel delay profile (eg, in Figure 6B or 6D) is determined based on the channel topology (eg, in Figure 6A or 6C) and the FFT start pointer. Conversely, the channel topology is determined based on the channel delay profile and the FFT start pointer.

Figures 6A-6D assume that the FFT start pointer is not in the useful part of the OFDM symbol. A given channel topology is also associated with a number of possible FFT start pointer locations. For example, the channel topology in Figure 6D is also obtained by the start of the FFT window in the Δ sample on the right side away from the start of the useful part of the OFDM symbol copy, that is, the G sample on the right side of the FFT start pointer location shown in Figure 6D. Will be done. Uncertainty at the FFT start pointer is also tested to determine the correct FAP and LAP.

The time tracking loop 320 attempts to keep the FFT start pointer within the ISI / ICI-free region. The time tracking loop 320 attempts to achieve this goal in the presence of timing fluctuations due to differences between clock frequencies in transmitters and receivers, fluctuations in clock frequencies, changes in communication channels, and so on. Accurate updates of the FFT start inter are essential to achieve good data detection performance.

Figure 7A shows a typical channel impulse response estimate in a receiver. The receiver has knowledge of the channel tap at each tap position, but does not know where the channel impulse response actually begins before time tracking. The FAP and LAP detector 412 in the time tracking loop 320 detects for FAP and LAP.

Figures 7B and 7C show two possible results for FAP and LAP detection for channel impulse response estimation shown in Figure 7A. The resulting channel delay profile in Figure 7B shows that the detected FAP is at exponent Δ and the detected LAP is at exponent Δ + Δ._hIndicates that it is in. The resulting channel delay profile in Figure 7C shows that the detected LAP is at exponent Δ and the detected FAP is at exponent Δ + Δ._hIndicates that it is in.

Based on the channel impulse response estimation shown in FIG. 7A, the FAP and LAP detector 412 provided the detected FAP and LAP, which delayed spread Δ as shown in FIGS. 7B and 7C._hDepends on the time tracking algorithm, the size and location of the channel tap, and perhaps beyond the limits of other factors. If the actual channel delay profile is as shown in Figure 7B and the FAPd and LAPd are correct, the channel topology shown in Figure 6A is based on the relationships between the channel topologies, the FFT start pointer, and the channel delay profile. It will be rebuilt. However, if the actual channel delay profile is as shown in Figure 7B and the FAPd and LAPd decisions are incorrect, the incorrect channel topology shown in Figure 6C will be reconstructed.

The FFT start pointer is updated differently for the channel topologies shown in Figures 6A and 6C. Therefore, misdetection of FAPs and LAPs leads to misinterpretation of the channel topology, which causes the FFT start pointer to be updated in a suboptimal or improper way, which is also for time tracking and data detection. Degrades the performance of both.

To resolve ambiguities in FAP and LAP, the receiver evaluates two assumptions about FAPd and LAPd decisions to determine whether these decisions are correct or incorrect. Assumption 0 represents the assumption that the FAPd and LAPd decisions are correct, and Assumption 1 represents the assumption that the FAPd and LAPd decisions are incorrect. For each assumption, as mentioned above in Figures 6A-6D, the receiver develops an OFDM symbolic structure and the channel topology for that assumption by utilizing the relationship between the FFT start pointer and the channel delay profile. To rebuild. From the reconstructed channel topology, the receiver can see where the guard interval and guard copy are located based on the OFDM symbol structure. The receiver then evaluates each assumption and correlates between the guard interval and the guard copy. Correct assumptions result in greater correlation and are used to correct incorrect FAPd and LAPd decisions.

Figures 8A-8D show examples for resolving ambiguities in FAP and LAP detection by correlating OFDM symbol data. For this example, the true FAP and LAP and the detected FAP and LAP are assumed to be as shown in Figure 8B, and the detected FAP and LAP are incorrect. In the following description, FAPh and LAPh represent the hypothetical FAP and LAP for a particular assumption, respectively.

For Assumption 0, as shown in Figure 8B, the Assumptions FAP and LAP are equal to the detected FAP and LAP, or FAPh0 = FAPd and LAPh0 = LAPd. The channel topology shown in Figure 8A is reconstructed based on the hypothetical FAP and LAP and FFT start pointers. For the examples shown in FIGS. 8A-8D, the correlation window covers from the first sample in the guard section of the OFDM symbol copy of FAP to the last sample in the guard section of the OFDM symbol copy of LAP. For assumption 0, the correlation window is the sample index T_{0, a} Begins with, and sample index T_{0, b} end with.

For Assumption 1, as shown in Figure 8D, the hypothetical FAPs and LAPs are equal to the detected FAPs and LAPs, respectively, or FAPh0 = LAPd and LAPh0 = FAPd. The channel topology shown in Figure 8C is reconstructed based on the assumed FAP and LAP and FFT start pointers shown in Figure 8D. For Assumption 1, the correlation window is the sample index T_{1, a} Begins with, and sample index T_{1, b} end with.

For each assumption i, the start T of the correlation window_{i, a}Is determined as follows: T_{i, a}= Start FFT + T_{FAP, i}-G (i = 0,1) Equation (8) However, T_{FAP, i}Is the tap location of FAPhi. T for assumption 0 shown in Figures 8A and 6B_{FAP, 0}= Δ + Δ_hAnd for Assumption 1 shown in Figures 8C and 6D, T_{FAP, 1}= Δ.

For each assumption i, the end of the correlation window T_{i, b}Ha: T_{i, b}= Start FFT + T_{LAP, i} (i = 0,1) Equation (9) Is determined as. However, T_{LAP, i}Is the tap location of LAPhi after causing either encapsulation effect. T for assumption 0 shown in Figures 8A and 8B_{LAP, 0}= Δ + G, where G is added to cause LAPh0, which appears to the left of FAPh0, by the encapsulation effect. T for Assumption 1 shown in Figures 8A and 6B_{LAP, 1}= Δ + Δ_hHere, LAPh1 appears on the right side of FAPh1 and has no enveloping effect, so G is not added.

The size of the correlation window for each assumption i is T_{i, b}-T_{i, a}Is calculated as. As shown in Figures 8A and 6C, the correlation windows for assumptions 0 and 1 generally have different sizes.

The correlation is made between the guard interval and the guard copy for each assumption i as follows:<maths num="5"><img file="JP4955693B2_D0005.tif" /></maths>

However, r_nIs the input sample at the sample index n of the received OFDM symbol, C_i Is the correlation result for assumption i, and "*" Represents a complex conjugate.

Figures 8A and 8C show guard sections as well as guard copies of each OFDM symbol copy. The guard section is indicated by gray shading, and the guard copy is indicated by a dotted box. For each OFDM symbol copy, the guard interval is a duplicate of the guard copy. As shown in equation (10), each sample in the guard interval is multiplied by the complex conjugate of the corresponding sample in the guard copy, and the multiplication result is cumulative over the length of the correlation window. (T in equation (10)_{i, b}-T_{i, a}) Explains the different correlation window sizes used for assumptions 0 and 1 and yields normalized correlation results.

If the hypothetical FAPs and LAPs are true FAPs and LAPs, as shown in Figure 8D, then the correlation window encloses a guarded interval and a guarded copy of the transmitted OFDM symbol, as shown in Figure 8C. The channel topology for true assumptions then leads to higher correlation results between guard intervals and guard copies. This is because these sections carry the same waveform from the transmitter. As shown in Figure 8B, if the hypothetical FAPs and LAPs are not true FAPs and LAPs, then the correlation window covers part of the actual guard interval and also part of the useful part of the transmission OFDM symbol. Channel topologies for false assumptions therefore result in lower correlations between "guard intervals" and "guard copies", which often contain statistically irrelevant random data. Correlation results for the two assumptions are thus used to determine which assumption is correct and to correct any incorrect FAPd and LAPd decisions resulting from using only channel impulse response estimates. Will be.

The correct assumption is determined as follows: C₀ C₁If, FAPo = FAPd and LAPo = LAPd, equation (11a) Soon C₁&gt; C₀If, FAPo = LAPd and LAPo = FAPd, equation (11b) However, FAPo and LAPo are the output FAPs and LAPs provided by unit 414 in FIG. Equation (11a) is when assumption 0 is correct and the detected FAPs and LAPs are provided directly as the output FAPs and LAPs, respectively. Equation (11b) is when Assumption 1 is correct and the detected FAPs and LAPs are exchanged and provided as the output FAPs and LAPs.

FIG. 9 shows various examples of the correlation window. For the examples shown in FIGS. 8A-8D, it is labeled as window 1 in FIG. 9, where the correlation window begins with the first sample of the guarded interval of the hypothetical FAP and ends with the last sample of the guarded interval of the hypothetical LAP. For the embodiment labeled as Window 2, the correlation window begins with the central sample of the guarded section of the hypothetical FAP and ends with the last sample of the guarded section of the hypothetical LAP. For the embodiment labeled as window 3, the correlation window covers the entire guard section of the hypothetical LAP. For the embodiment labeled as window 4, the correlation window covers part of the entire guard section of the hypothetical LAP (eg, the second half). For the embodiment labeled as window 5, the correlation window covers the entire guard section of the hypothetical FAP. For the embodiment labeled as window 6, the correlation window covers the ISI / ICI region of the hypothetical FAP and LAP. For the embodiment labeled as window 7, the correlation window has a fixed size (eg, G sample) and is centered in the ISI / ICI region of the hypothetical FAP and LAP. Various other correlation windows are also used. In another embodiment not shown in FIG. 9, the correlation window size is W max {Δ_h} Is selected. In this example, the previous FAP / LAP location is used to limit the correlation window size.

FIG. 10 shows an example of process 1000 for resolving ambiguity in the detected FAP and LAP. FAPs and LAPs are initially detected, for example, based on channel impulse response estimates for communication channels (block 1012). FAP and LAP determine, for example, the energy of channel taps in the sliding window for different tap positions, as described above for equations (1)-(7), and a finite difference value based on that energy. Is detected by detecting FAP based on the first function of energy and finite difference value, and detecting LAP based on the second function of energy and finite difference value.

Assumptions regarding the first assumption corresponding to correctly detected FAPs and LAPs FAPs and LAPs are determined based on the detected FAPs and LAPs (block 1014). Assumptions regarding the second assumption corresponding to falsely detected FAPs and LAPs FAPs and LAPs are also determined based on the detected FAPs and LAPs (block 1016). The correlation window is determined for each assumption based on the assumptions FAP and LAP for that assumption (block 1018). The correlation window for each assumption covers all or part of the guarded section of the hypothetical FAP and / or all or part of the guarded section of the hypothetical LAP.

Correlation is then performed on the received data on the first assumption based on the correlation window on the first assumption (block 1020). Correlation is also performed on the received data for the second assumption based on the correlation window for the second assumption (block 1022). For each assumption, a correlation is performed between the first segment of the received data in the correlation window for that assumption and the first segment of the received data that is K samples away, where K is the useful part of the OFDM symbol. The period. The correct assumptions are then determined based on the correlation results for the first and second assumptions (block 1024).

In another form, the correct FAP and LAP are determined by evaluating one assumption. FAP and LAP are detected first as mentioned above. One assumption regarding FAP and LAP is selected for evaluation in various ways.

In one embodiment, assumptions are selected for evaluation that result in a small timing adjustment to the current FFT start pointer. In this example, the amount of timing adjustment to the current FFT start pointer for Assumption 0 is determined, and ΔT.₀It is expressed as. The amount of timing adjustment to the current FFT start pointer for Assumption 1 is also determined, and ΔT₁It is expressed as. Assumptions are selected for evaluation by smaller timing adjustments, and are called assumptions s. Correlation is performed for the selected assumptions s as described above. Correlation result for assumption s C_s Is the threshold C_thIs compared against. C_s Is C_thIf greater, the assumption s is considered to be the correct assumption, and the FFT start pointer is updated based on the assumptions FAP and LAP for the assumption s. In other cases, C_s Is C_thIf equal to or less than, the other unevaluated assumptions (called assumptions u) are considered correct assumptions, and the FFT start pointer is updated based on the assumptions FAP and LAP for assumption u.

In another embodiment, the assumptions are selected for evaluation based on the timing (eg, timing adjustment) with respect to the previous assumption determination. In this embodiment, the timing adjustment is calculated for assumptions 0 and 1 as described above. The timing adjustments for the two assumptions are compared to the timing adjustments for the correct assumptions for the previous update interval. For example, timing adjustments for the P newest correct assumptions are stored, and assumptions 0 or 1 closer to more of the stored P assumptions are selected for evaluation. As another example, assumptions 0 or 1 that are close to the average timing of the stored P assumptions are selected. Other forms of history information are also stored and used for assumption selection. In any case, the selected assumptions are evaluated and the correlation result C to determine if the selected assumptions s or the unselected assumptions u are correct assumptions._sIs the threshold C_th Is compared against.

As mentioned above, one assumption is selected for evaluation based on the detected FAP and LAP and background information. One assumption is also selected without background information. For example, assumption 0 is always selected, or one assumption is randomly selected. The selected assumptions are evaluated and the threshold C_thIs compared against. Threshold C_thIs a constant value selected to achieve good performance. Threshold C_thIs also configurable, for example, set based on the correlation results for recent correct assumptions. The correlation window for the selected assumptions is determined based on one of the examples described above for FIG. Variable size correlation windows improve detection performance. However, constant size correlation windows reduce the complexity of implementation.

FIG. 11 shows an example of process 1100 for resolving ambiguity in detected FAPs and LAPs. FAPs and LAPs are initially detected, for example, based on channel impulse response estimates for communication channels (block 1112). One assumption regarding FAP and LAP is selected for evaluation (block 1114). Assumptions are selected based on the current FFT start pointer, background information on previous assumptions, and so on. Correlation is performed on the received data for the assumptions selected there (block 1116). Whether the selected assumption is correct is determined based on the correlation results and thresholds (block 1118).

In the technique described here, when the channel delay spread is less than half of the guard section, the channel delay spread is also more than half of the guard section (or Δ)._hWhen&gt; G / 2), it is used to resolve ambiguities in FAP and LAP detection. Generally, the technique is used to resolve ambiguity in either two channel taps for the impulse response of a communication channel, where the ambiguity is faster for the first channel tap than for the second channel tap. Due to uncertainty about whether it is slow or not. Correlation is performed on the first assumption corresponding to the first channel tap, which is earlier than the second channel tap. Correlation is also performed for the second assumption corresponding to the first channel tap, which is slower than the second channel tap. Correlation results for the two assumptions are used to determine if the first channel tap is faster or slower than the second channel tap.

With reference back to FIG. 4, the centroid detector 416 receives the output FAP and LAP from the unit 414 and determines the centroid of the channel impulse response estimate. In the embodiment, the detector 416 determines the center of gravity solely based on the output FAP and LAP. For example, the center of gravity is set equal to the midpoint between FAPo and LAPo as follows: DS = (LAPo-FAPo) mod N, equation (12a) CM = (FAPo + DS / 2) mod N, equation (12b) However, DS is the detected channel delay spread. In equation (12a), the channel delay spread is calculated as {(LAPo-FAPo) mod N}, where the mod N operation describes the case where LAPo is to the left of FAPo in the channel delay profile. In equation (12b), the centroid is half the channel delay spread from FAPo, and here we describe the inclusion of the centroid in which the mod N operation is possible. In another embodiment, the centroid is determined based on two or more (eg, all) channel taps, strong enough channel taps, etc. in the channel impulse response estimation. The centroid calculation is performed by other methods known in the art.

The timing error from adder 418 is calculated as follows: Timing error = CM-Timing target equation (13) Timing errors are signed values that are positive, zero, or negative.

In one embodiment, the timing adjustment from loop filter 420 is calculated as follows: Timing adjustment (t) = Timing error x Gain + Timing adjustment (t-1) Equation (14) In another example, the timing adjustment is calculated as: Timing adjustment (t) = Timing error x Gain 1+ Timing adjustment (t-1) x gain 2 equation (15) The gain in Eq. (14) and the gain 1 and gain 2 in Eq. (15) are fixed or programmable values.

Timing adjustments are limited to a given value range as follows: Adv / Ret (t) = SAT {Timing adjustment (t), maximum adjustment} Equation (16) However, Max Adjust is the absolute value for maximum timing adjustment for any given update, and SAT {} is a saturation operation. The resulting Adv / Ret (t) is a signed saturation value.

The FFT start pointer is updated as follows: FFT start pointer (t) = FFT start pointer (t-1) + Adv / Ret (t) Equation (17) The FFT start pointer is advanced for each OFDM symbol by the K + G sample period, and is further updated by Adv / Ret (t) at each loop update interval.

The time tracking loop 320 attempts to maintain the center of gravity at a specific location in the channel delay profile, which is referred to as target timing. A good location for target timing depends on the actual channel impulse response, the length of the channel estimation window (N), the length of the guard interval (G), and so on. Programmable values are used for target timing. Loop filter gain, Max Adjust and / or other parameters are fixed or configurable values.

FIG. 12 shows an example of the process 1200 for performing time tracking in the receiver. FAPs and LAPs are initially detected, for example, based on channel impulse response estimates for communication channels (block 1212). Ambiguity in the detected FAP and LAP is then resolved by performing a correlation on the received data for at least one assumption about the FAP and LAP (block 1214). One assumption is evaluated as described in FIG. 10, or one assumption is evaluated as described in FIG. Timing at the receiver is then updated based on the assumptions FAP and LAP for the correct assumptions (block 1216). For block 1216, the centroid of the channel impulse response estimate is determined based on the assumptions FAP and LAP on the correct assumptions (eg, as the midpoint between them). Timing errors are determined based on the center of gravity and timing goals, and are filtered to obtain timing adjustments. Timing adjustments are limited to a range of predetermined values and are used to update receiver timing, eg, the FFT start pointer. OFDM demodulation is performed on the received data based on the updated receiver timing (block 1218). The FFT start pointer is used to select a sample in the received data for processing.

The techniques described herein are implemented by a variety of means. For example, these techniques are implemented in hardware, firmware, software, or a combination thereof. For hardware implementation, the processing units used to resolve ambiguities in channel estimation and / or to perform time tracking in the receiver are one or more application specific integrated circuits (ASICs), digital signal processors (DSPs). ), Digital Signal Processing Devices (DSPDs), Programmable Logical Devices (PLDs), Field Programmable Gate Arrays (FPGAs), Processors, Controllers, Microcontrollers, Microprocessors, Electronic Devices, Designed to Perform the Functions Described Here It is carried out in other electronic units, or a combination thereof.

For firmware and / or software implementation, the technology is implemented by modules (eg procedures, functions, etc.) that perform the functions described herein. The software code is stored in memory (eg, memory 182 in FIG. 1) and executed by a processor (eg, processor 180). Memory is implemented inside or outside the processor.

The above description of the Disclosure Examples is provided to allow one of ordinary skill in the art to make or use the present invention. Various modifications to these embodiments will be immediately apparent to those skilled in the art, and the general principles defined herein apply to other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not intended to be limited to the examples presented herein, and should be given the broadest scope compatible with the principles and novel features disclosed herein.

<figref num="1">The block diagram of the transmitter and the receiver is shown.</figref><figref num="2">The block diagram of the OFDM modulator of the transmitter is shown.</figref><figref num="3">The block diagram of the OFDM demodulator of the receiver is shown.</figref><figref num="4">The block diagram of the time tracking loop in the receiver is shown.</figref><figref num="5A">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="5B">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="5C">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="5D">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="5E">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="5F">Illustrates FAP and LAP detection for two typical channel impulse response estimates.</figref><figref num="6A">The channel topology and channel impulse response for two typical operating scenarios are shown.</figref><figref num="6B">The channel topology and channel impulse response for two typical operating scenarios are shown.</figref><figref num="6C">The channel topology and channel impulse response for two typical operating scenarios are shown.</figref><figref num="6D">The channel topology and channel impulse response for two typical operating scenarios are shown.</figref><figref num="7A">The possible decisions made by the FAP and LAP detectors are shown.</figref><figref num="7B">The possible decisions made by the FAP and LAP detectors are shown.</figref><figref num="7C">The possible decisions made by the FAP and LAP detectors are shown.</figref><figref num="8A">An example is illustrated for resolving ambiguity in FAP and LAP detection by correlating OFDM symbol data.</figref><figref num="8B">An example is illustrated for resolving ambiguity in FAP and LAP detection by correlating OFDM symbol data.</figref><figref num="8C">An example is illustrated for resolving ambiguity in FAP and LAP detection by correlating OFDM symbol data.</figref><figref num="8D">An example is illustrated for resolving ambiguity in FAP and LAP detection by correlating OFDM symbol data.</figref><figref num="9">Shows the correlation window used to resolve FAP and LAP ambiguities.</figref><figref num="10">The process for resolving the ambiguity in the detected FAP and LAP is shown.</figref><figref num="11">Here is another action to resolve the ambiguity in the detected FAPs and LAPs.</figref><figref num="12">The process for performing time tracking in the receiver is shown.</figref>