RU2373660C2 - Coherent tracking for frequency-modulated receiver of wpd with help of antenna system with switched diversity - Google Patents

Abstract

FIELD: radio engineering.

SUBSTANCE: invention is related to the field of radio signals reception and may be used in reception of signals with the help of antenna system with switched diversity. Method for coherent tracking of radio signal, including multiple reference subcarriers exposed to digital modulation, includes stages, at which symbols sent on reference subcarriers are received, symbols of reference subcarriers are combined with conjugate value of available reference sequence for provision of multiple samples, median filtration of samples is done to obtain filtered samples, and samples are smoothened for each of reference subcarriers on multiple reference subcarriers to obtain coherent estimate of reference signal for each of subcarriers.

EFFECT: provision of matching for transitional switching processes and improved algorithms of coherent tracking.

32 cl, 52 dwg

Description

FIELD OF THE INVENTION

This invention relates to methods and devices for receiving radio signals, and more particularly to such methods and devices that include an antenna diversity switching system.

State of the art

HD Radio ^TM system from iBiquity Digital Corporation is designed to provide the sound quality of the digital level, superior to existing analog broadcasting formats. The HD Radio ^{TM system} allows a gradual evolution from modern broadcasting systems using amplitude modulation (AM broadcasting) and broadcasting using frequency modulation (FM broadcasting) to a fully digital system of standard in-band per-channel broadcasting (VOC (IBOC)). This system provides audio and data transmission services to mobile, portable and fixed receivers from terrestrial transmitters in the existing medium frequency (MF; MF) and very high frequency (VHF; VHF) radio bands. Broadcasters can continue AM and FM broadcasting simultaneously with new, higher quality and more robust digital signals, which allows these companies and their listeners to switch from analogue to digital broadcasting, while maintaining their current frequency assignments.

HD Radio ^TM broadcast in accordance with the VLOOKUP standard is performed using a composite signal, which includes many subcarriers obtained by orthogonal frequency division multiplexing (OFDM), and reference subcarriers within the broadcast channel. Coherent demodulation is used for the digital part of the VLOOKUP FM signal in HD Radio ^TM standard receivers. Numerous functions performed by reference subcarriers to collect, track, and evaluate channel state information (CSI) and coherent operation are described in US Pat. No. 6,549,544. The system described in US Pat. No. 6,549,544 is designed to operate in a frequency band ( 88-108 MHz) reserved for FM broadcasting with fading in bandwidth for matching with receivers used in vehicles traveling fast on highways. Various coherent tracking parameters are evaluated using filters whose bandwidths approach the maximum expected bandwidth (approximately 13 Hz) of Doppler frequencies. With a stationary antenna, relevant tracking statistics for the input signal for tracking algorithms are assumed to vary in frequency not exceeding the Doppler frequency bandwidth.

Receivers HD Radio ^TM CDF standard can be used in conjunction with the antenna system with switch diversity. The antenna diversity switching system includes several (for example, from 2 to 4) antenna elements, which are usually enclosed inside the glass of the front or rear windows of the vehicle. These elements are connected to a diversity switching module that dynamically selects one element or combination of elements to provide a radio frequency (RF) antenna signal to the receiver. The diversity switching module also operatively monitors the signal from the receiver to determine when switching is needed. A typical “blind switching” algorithm sets the switching threshold based on the average level of the intermediate frequency (IF) signal from the receiver. When the IF signal falls below this threshold, the switch blindly selects an alternate element from which it is expected to produce the best signal. If the new signal exceeds the threshold, the switch supports the selection of a new element. Otherwise, the diversity switching module selects an alternative element after a minimum amount of a delayed operation has expired. This process continues, and the diversity switching module continuously updates its threshold (its thresholds). An example of such a switched diversity antenna system is presented in US Pat. No. 6,633,258 B2 (H. Lindenmeier et al.) “Diversity System for Receiving Digital Terrestrial and / or Satellite Radio Signals for Motor Vehicles” (“A diversity reception system for radio signals transmitted from ground stations or satellites, for cars "), issued on October 14, 2003

The theoretical rationale for the operation of the diversity switching algorithm is based on different instantaneous fading characteristics of various antenna elements. Multipath fading causes an increase in the multiple beams (multipath) of the signal arriving at the receiving antenna element at different points in time. For example, a wavelength at 100 MHz is approximately 10 feet. If two signal propagations arrive at a point in time that differs from the corresponding one at the same wavelength, i.e. 10 nanoseconds (10 ft. Travel difference), these signals will add up in phase. Similarly, if two beams arrive at the antenna element with a time difference corresponding to the half-length of the wave, then these signals folding in antiphase will be mutually destroyed. This addition or mutual annihilation is dynamic in a moving vehicle, where the Doppler frequency bandwidth is approximately determined by the expression BW = f _c · s / c (where f _c is the carrier frequency, s is the vehicle speed, c is the speed of light). The Doppler frequency bandwidth is approximately 10 Hz at speeds typical of freeways. Therefore, the signal vector (complex version of the module and phase) of one antenna element can change with a frequency of about 10 Hz in this example. Then the coherent tracking of the reference signal and the channel state must be consistent with a bandwidth of 10 Hz in order to maintain coherent tracking of the signal.

Typical antenna elements in a vehicle may be subject to some instantaneous fading conditions (depending on the gap between the elements and the directions of multipath propagation). For example, one element may be at zero fading, and another element at maximum. In a vehicle with several such elements, it is likely that some alternative element will receive a significantly larger signal when the given element experiences fading (mutual annihilation of signals). Typical elements in a multi-element diversity FM antenna system will have instantaneous fading conditions that could be made correlated to some extent, but they are left uncorrelated enough to achieve the desired diversity gain to improve performance.

The coherent digital modem in the IBOC HD Radio ^TM receiver is designed to track signal fading at vehicle speeds where the Doppler frequency bandwidth is less than 13 Hz. The use of switched diversity antennas in vehicle windows introduces fast transients into coherent tracking of a digital signal, which degrades the performance of digital broadcasting. It is impossible to keep track of transients caused by dynamic switching of antennas with the help of the previously developed modem, which leads to a reduced service area for digital broadcasting.

In this invention, a coherent tracking method is provided that provides coordination of switching transients in a switched diversity antenna system. It also provides improvements to coherent tracking algorithms that are used autonomously in a blind switching antenna system. The same algorithms can also neutralize performance degradation due to impulse noise or non-Gaussian noise, for example, from a neighboring analogue source of FM interference. In addition, the performance of receivers that provide fast automatic gain control (AGC) is improved.

Summary of the invention

A method for coherent tracking of a radio signal including a plurality of reference subcarriers subjected to digital modulation is proposed. The method includes the steps of receiving the symbols transmitted on the reference subcarriers, combining the symbols of the reference subcarriers with the conjugate value of the known reference sequence to obtain a plurality of samples, median filtering of the samples to obtain filtered samples, and smoothing the samples for each of the reference subcarriers according to the set of reference subcarriers to obtain a coherent estimate of the reference signal for each of the subcarriers.

In yet another aspect, the invention describes a receiver for coherently tracking a radio signal including at least one reference carrier digitally modulated. The receiver comprises an input for receiving a radio signal and a processor for coherently tracking a radio signal including a plurality of reference subcarriers subjected to digital modulation by receiving symbols transmitted on the reference subcarriers, combining the symbols of the reference subcarriers with the conjugate value of the known reference sequence to obtain a plurality of samples, median filtering of samples to obtain filtered samples and smoothing samples for each of the reference subcarriers across multiple reference subcarriers to obtain a coherent estimate of the reference signal for each of the subcarriers.

The invention also describes a method for coherently tracking a radio signal including at least one reference carrier, digitally modulated, including the steps of demodulating the reference carrier to obtain complex coherent reference gain factors, detecting a transient process that affects to complex coherent reference gain, and adjust the complex coherent reference gain in the vicinity of the transient to get adjusted complex reference gains. The transient can be caused by switching between antenna elements or impulse noise.

The invention also describes a receiver for coherent tracking of a radio signal including at least one reference carrier, subjected to digital modulation. The receiver comprises an input for receiving a radio signal and a processor for demodulating the reference carrier to obtain complex coherent reference gain factors, to detect a transient process that affects the complex coherent reference gain factors, and to control complex coherent reference gain factors in the vicinity of the transition process to obtain adjusted coherent reference gain factors.

In yet another aspect, the invention provides a method for estimating a noise variance of symbols in a radio signal, where the noise may include pulsed noise samples among Gaussian noise samples, the method including the steps of summing input samples and coherent reference samples to obtain error samples squares of error samples are computed, squared samples of Gaussian noise and squared samples of impulse noise are divided, and non-linear filtering of squares of samples is performed errors to obtain an estimate of the noise variance representing the sum of long-term-averaged Gaussian-like noise variance, and short-term impulsive noise variance.

Brief Description of the Drawings

Figure 1 shows a schematic representation of the spectrum of a hybrid signal to which this invention can be applied.

Figure 2 shows a schematic representation of the spectrum of an extended hybrid signal to which this invention can be applied.

Figure 3 shows a schematic representation of the spectrum of a fully digital signal to which this invention can be applied.

Figure 4 shows a schematic representation of a first type of ordering of frequency segments.

Figure 5 shows a schematic representation of a second type of ordering of frequency segments.

Figure 6 shows a schematic representation of the spectral display of the reference subcarriers of the lower side strip.

7 is a schematic representation of a spectral display of reference subcarriers of the upper side strip.

Fig. 8 is a block diagram of a system control processing module.

FIG. 9 is a schematic representation of a sequence of control data of a system of primary reference subcarriers.

Figure 10 presents a block diagram of a differential encoder.

Figure 11 presents a high-level block diagram of the functional module of the coherent reference signal and CSI (IOSK).

On Fig presents a detailed high-level block diagram of the functional module of the coherent reference signal and IOSK.

13 is a schematic representation of data deletion from reference subcarriers of the upper primary sideband.

On Fig presents a block diagram of the assessment of the reference subcarriers of the primary side bands.

On Fig presents a block diagram of the assessment of the reference subcarriers of the secondary side bands.

On Fig presents a block diagram of the estimation of the variance of the noise of the primary side bands.

17 is a flowchart for estimating the noise variance of the secondary sidebands.

On Fig presents a block diagram of the weighing IOSK reference subcarriers.

On Fig presents a block diagram of the generation of weights synchronization.

FIG. 20 is a schematic illustration of an illustration of interpolation of reference subcarriers for primary sidebands.

On Fig presents a block diagram of the interpolation of the reference subcarriers.

On Fig presents a functional diagram of a characteristic antenna system with switched diversity.

On Fig presents a graph of the real and imaginary components of the signal fading (no noise) in the case of switching every 100 characters.

On Fig presents a graph of the real and imaginary components of the filtered coherent reference fading signal in the vicinity of the transient on the symbol at number 200 with a signal-to-noise ratio (SNR) of 10 dB.

25 is a graph illustrating the effects of switching transients on noise estimates (SNR 20 dB).

On Fig presents a block diagram of a preliminary coherent estimate of the reference signal and noise.

On Fig presents a graph in the case of an 11-tap filter with a finite impulse response (FIR filter) for coherent estimation of the reference signal of a single reference subcarrier.

On Fig presents a graph in the case of using a frequency filter (to obtain reference subcarriers).

On Fig presents a graph of the effects of the type of filter to reduce the estimation error for a coherent reference signal.

On Fig presents a graph comparing the characteristics of the 11-tap FIR (FIR) filter and median filtering of the coherent reference signal.

On Fig presents a block diagram of the evaluation of the reference signal and noise according to the present invention.

On Fig presents a graph of the impulse response of a filter with an infinite impulse response (IIR filter).

On Fig presents a graph of the response of the IIR filter to the jump.

On Fig presents a graph of the reduction of errors in the estimation of noise as a function of the parameter "a" IIR filter.

Fig. 35 is a graph illustrating the effects of an excess gain value G ∗ ln (2) on the probability of choosing an excess path and on the increased average estimated noise variance.

36 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) with the original 11-tap FIR filter and the original IIR noise estimate with a = 1/16.

Fig. 37 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) with the original 11-tap FIR filter and the IIR noise estimate with change, according to which a = 1/8.

38 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) with the original 11-tap FIR filter when a variation in noise variance is superimposed.

Fig. 39 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) with a 5-tap median filter and IIR noise estimate with variation, according to which a = 1/8.

40 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) with a 5-tap median filter when a change in noise variance is superimposed.

On Fig and 42 presents graphs of the frequency of occurrence of errors in blocks.

On Fig presents a block diagram of a receiver that can process signals in accordance with the invention.

Fig. 44 is a flowchart illustrating a channel state estimation method used in the receiver of Fig. 43.

Fig. 45 is a graph illustrating the effects of switching transients on noise estimates.

Fig. 46 is another graph illustrating the effects of switching transients on noise estimates.

On Fig presents a functional block diagram illustrating the generation of coherent reference channel signals.

48 is a graph illustrating the results of a transient detection algorithm.

49 is another graph illustrating the results of a transient detection algorithm.

On Fig presents a functional block diagram illustrating the correction of the gain of the coherent channel in the vicinity of the transition.

On Fig presents a block diagram illustrating a modification of the estimation of noise variance.

52 is a graph illustrating the effects of switching transients on noise estimates.

DETAILED DESCRIPTION OF THE INVENTION

Figure 1 shows a schematic representation of the spectrum of a hybrid FM signal 50 of the IBOC standard, to which this invention can be applied. This signal includes an analog modulated signal 52 located in the center of the broadcast channel 54, a first plurality of equally spaced subcarriers 56, orthogonal frequency division multiplexed, in the upper sideband 58, and a second plurality of equally spaced subcarriers 60, subjected to orthogonal frequency division multiplexing (OFDM) in the lower sideband 62. Digitally modulated subcarriers are transmitted at a lower power level than the analog mode ulirovannaya carrier, to align with the required channel signal masks. Digitally modulated subcarriers are segmented, and various subcarriers are designated as reference subcarriers. The frequency segment is a group of 19 subcarriers subjected to OFDM (OFDM subcarriers) and containing 18 subcarriers, which are data subcarriers, and one reference subcarrier.

The hybrid signal includes an analog frequency modulated (FM) signal and primary primary subcarriers subjected to digital modulation. The digital signal is transmitted in primary primary (PeO) sidebands on either side of the analog FM signal in the hybrid signal. The power level of each sideband is noticeably lower than the full power of the analog FM signal. An analog signal can be monaural or stereo and can include channels that comply with the Subsidiary Communications Authorization (SCA) Regulation (FDA) for FM broadcasting in sound and amplification systems).

Subcarriers are located in equally spaced parts of the frequency spectrum. Subcarrier locations are numbered from -546 to +546. In the signal shown in FIG. 1, the subcarriers are located at locations numbered +356 to +546 and -356 to -546. This signal will usually be used in the initial transition phase before being converted to a fully digital signal.

Each primary main sideband consists of ten frequency segments, which are distributed among subcarriers numbered from 356 to 545 or from -356 to -545. Subcarriers 546 and -546, also included in the primary main sidebands, are additional reference subcarriers. The amplitude of each subcarrier can be scaled using a scale factor of the amplitude.

In a hybrid signal, the digital signal is transmitted in the primary main (PeO) sidebands on either side of the analog FM signal. The power level of each sideband is noticeably lower than the full power of the analog FM signal. An analog signal may be monophonic or stereo and may include channels that comply with the regulations of the VTS.

Figure 2 shows a schematic representation of the spectrum of an extended hybrid signal to which this invention can be applied. In an extended hybrid signal, the sideband of the hybrid signal can be expanded to an analog FM signal to increase digital capacitance. This additional spectrum, assigned to the inner edge of each primary main sideband, is called the primary widened (PRa) sideband.

An extended hybrid signal is created by adding primary extended sidebands to the primary main sidebands present in the hybrid signal, as shown in FIG. Depending on the service mode, one, two or four frequency segments can be added along the inner edge of each primary main sideband.

Figure 2 shows a schematic representation of an extended hybrid FM signal 70 standard VLOOKUP. This extended hybrid signal is created by adding the primary extended sidebands 72, 74 to the primary primary sidebands present in the hybrid signal, as shown in FIG. Depending on the service mode, one, two or four frequency segments can be added along the inner edge of each primary main sideband.

The hybrid signal includes an analog frequency-modulated (FM) signal and primary primary subcarriers, digitally modulated (subcarriers numbered +356 to +546 and -356 to -546) and several primary extended subcarriers (subcarriers numbered + 280 to +355 and from -280 to -355) or all primary extended subcarriers. This signal will usually be used during the initial transition phase prior to conversion to a fully digital signal.

Each primary main sideband includes ten frequency segments and an additional reference subcarrier spanning subcarriers numbered 356 to 546 or -356 to -546. The upper primary extended sidebands include subcarriers numbered from 337 to 355 (one frequency segment), from 318 to 355 (two frequency segments), or from 280 to 355 (four frequency segments). The lower primary extended sidebands include subcarriers numbered from -337 to -355 (one frequency segment), from -318 to -355 (two frequency segments), or from -280 to -355 (four frequency segments). The amplitude of each subcarrier can be scaled using a scale factor of the amplitude.

Figure 3 shows a schematic representation of the spectrum of a fully digital signal to which this invention can be applied. Figure 3 shows a schematic representation of a fully digital FM signal 80 of the standard VLOOKUP. This fully digital signal is created by inhibiting the analog signal, completely expanding the bandwidth of the primary digital sidebands 82, 84 and adding secondary sidebands 86, 88 of lower power to the spectrum freed by the analog signal. The fully digital signal in the illustrated embodiment includes digitally modulated subcarriers at locations of subcarriers numbered from -546 to +546, without an analogue FM signal.

In addition to the ten main frequency segments, all four extended frequency segments are present in each main sideband of a fully digital signal. Each secondary sideband also has ten secondary primary (WO) and four secondary extended (BPa) frequency segments. However, unlike the primary sidebands, the secondary primary frequency segments are displayed closer to the center of the channel, and the expanded frequency segments are displayed farther from this center.

Each secondary sideband also supports a small secondary protected (WZ) region 90, 92, including 12 OFDM subcarriers and reference subcarriers 279 and -279. These sidebands are called “protected” because they are in the spectral region where the likelihood of analog or digital interference is least. An additional reference subcarrier is located in the center of the channel (0). The ordering of the frequency segments of the WZ region is not applied, because the WZ region does not contain frequency segments.

Each secondary primary sideband contains subcarriers numbered from 1 to 190 or from -1 to -190. The upper secondary sideband includes subcarriers numbered 191 to 266, and the upper secondary protected sideband includes subcarriers numbered 267 to 278 and the reference subcarrier numbered 279. The lower secondary sideband includes subcarriers numbered from - 191 to -266, and the lower secondary protected sideband includes subcarriers numbered -267 to -278 and an additional reference subcarrier numbered -279. The total frequency extent of the fully digital spectrum is 396.803 Hz. The amplitude of each subcarrier can be scaled using a scale factor of the amplitude. Scale amplitude coefficients of the secondary sidebands can be selected by the user. For application to the secondary sidebands, any of four factors can be selected.

All three types of signals considered correspond to the currently selected mask of spectral radiation. The digital signal is modulated by orthogonal frequency division multiplexing (OFDM). OFDM is a parallel modulation procedure in which a data stream modulates a large number of orthogonal subcarriers that are transmitted simultaneously. OFDM has the flexibility of easily mapping logical channels to different groups of subcarriers.

OFDM subcarriers are collected in frequency segments. Figure 4 shows a schematic representation of a first type of ordering of frequency segments. Figure 5 shows a schematic representation of a second type of ordering of frequency segments. Each frequency segment consists of eighteen data subcarriers and one reference subcarrier, as shown in FIG. 4 (ordering A) and FIG. 5 (ordering B). The position of the reference subcarrier (ordering A or ordering B) changes with the location of the frequency segment within the spectrum.

Signal processing is carried out at the protocol levels in the transmitter of the digital broadcasting system for transmitting audio information. Control and information signals are passed through various levels of the protocol stack to generate a VLOOKUP signal on the broadcasting side. Receivers process this signal at the appropriate protocol levels.

The Main Program Service stores the existing programming formats for analogue radio signals in both analogue and digital transmission. In addition, the main software service may provide for the availability of digital data that directly correlates with the programming of audio information.

Between peer layers (for example, from level n on the transmitting side to level n on the receiving side), a protocol data unit (PDU) is exchanged. The fundamental goal of any level n of the protocol stack is to deliver protocol data blocks issued by transmitter level n + 1 to the equal level n + 1 of the receiving system. The payload of an n + 1 PDU consists of protocol control information (PCI) of an n + 1 level and an upper layer PDU (an n + 2 level).

To better understand this idea, consider the flow of information from level n + 1 to level n on the transmitting side. Protocol data blocks of level n + 1 shall be packed as specified by the level n service. This package is called the service data unit or SDU. Level n DBMS includes n + 1 level BDBs and SBD control information (DBMS). Level n + 1 creates an SBU of level n and sends this block to level n through an access point to a service of level n.

When level n receives an SDB, it takes a n + 1 PDU and its own protocol control information (SIP), which may include information received in the UBIS, and creates a level n PDB. Then, the n-level PDUs are sent to the peer in the receiving system, where the process proceeds essentially in the reverse order when the information goes up the protocol layers. Therefore, each level has its own equal PBB and forwards the remaining information to the next level in the form of an SDB.

L1 is a physical layer in which signals are prepared for transmission through an antenna. L2 level is associated with L1 level. For each frequency segment, data subcarriers d1-d18 transport protocol data units of L2 level, and reference subcarriers transport system level L1 control information. The subcarriers are assigned numbers from 0 at the center frequency to ± 546 at one or the other end of the channel frequency distribution.

In addition to the reference subcarriers within each frequency segment, depending on the service mode, up to five additional reference subcarriers are introduced into the spectrum, and the numbers of these subcarriers are as follows: -546, -279, 0, 279, 546. The overall effect is the regular distribution of reference subcarriers throughout the spectrum. For convenience of notation, each reference subcarrier is assigned a special identification number between 0 and 60. All reference subcarriers of the lower side bands are shown in FIG. All reference subcarriers of the upper side bands are shown in FIG. In these drawings, the relationship between the numbers of the reference subcarriers and the numbers of OFDM subcarriers is indicated.

1-3 show the number of subcarriers and the center frequency of some key OFDM subcarriers. The center frequency of the subcarrier is calculated by multiplying the number of the subcarrier by the gap Δf between the OFDM subcarriers. The center of the subcarrier at number 0 is at a frequency of 0 Hz. In this context, the center frequency is relative to the allocated RF channel, for example, the upper primary main sideband is limited to subcarriers 356 and 546, the center frequencies of which are located at 129.361 Hz and 198.402 Hz, respectively. The frequency length of the primary primary sideband is 69.041 Hz (= 198.402-129.361).

Fig. 8 is a block diagram of a system control processing module. The system control channel (SCCH) transmits control and status information using fields in the reference subcarriers. In addition, several bits of the system control data sequence, called "backup", are controlled from levels above L1 through the primary interface of reserved control data and secondary interface of reserved control data Maintenance modes prescribe all valid logical channel configurations.

Under the control of higher levels, the system control processing module assembles and differential encodes a sequence of binary bits (sequence of system control data) intended for each reference subcarrier, as shown in FIG. In total, there are up to sixty one reference subcarriers numbered from 0 to 60, and these subcarriers are distributed throughout the spectrum where the OFDM is performed.

As shown in FIG. 8, the system control processing module receives input signals from level 2 through the CSC. Using these system control input signals, the system control data collection module 100 creates a sequence of system control binary bits of length T _b for each of the reference subcarriers, the number of which is 61. Then, the differential encoder 102 differentially encodes each binary sequence at a time. The resulting output signal is a 32 × 61 fixed-size matrix R. The dimension of the rows of the matrix R corresponds to the number of OFDM symbols per T _b , and the dimension of the columns corresponds to the maximum number of active reference subcarriers per OFDM symbol. The matrix R is available for the display module OFDM subcarriers with a speed of R _b . In addition, the system control processing module provides a count of blocks of level L1 to level 2 with a speed R _b through the control panel.

The module for collecting system control data collects all the control information of the system from level 2 and, composing it together with the control information of a certain level, builds a matrix r consisting of 61 sequences of 32-bit system control data. The rows of the matrix r are numbered 0 ... 31, and the columns are numbered 0 ... 60. Each row of the matrix r contains one binary bit of the system control data sequence for each reference subcarrier (before differential encoding) and is transmitted in the same OFDM symbol. Line number 0 is filled first. Any given column of matrix r contains a sequence of system control data for one reference subcarrier with a length of 32 OFDM symbols.

FIG. 9 is a schematic representation of a sequence of 104 control data of a primary reference subcarrier system. This sequence of system control data consists of binary fields, which are various components of the system control. The reference subcarriers located in the primary sidebands have fields different from those that have the reference subcarriers located in the secondary sidebands. Information contained in primary reference subcarriers applies only to primary services. The sequence of system control data by means of primary reference subcarriers is shown in Fig. 9 and is described in Table 1. Binary bits from the 31st to zero are respectively mapped to rows from zero through the 31st matrix r . Secondary reference subcarriers are characterized in a similar manner.

Table 1
Primary System Control Data Sequence Binary Map Field Binary Numbers Binary Length Description Fourth to tenth synchronization bits 31:25 7 Fourth to tenth synchronization bits = 0110010 Second backup binary bit 24 one Managed by higher layers of the protocol stack Third Binary Parity 23 one Parity Check for Backup Binary Third Binary Timing Bit 22 one Third Binary Sync Bit = 1 Zero and first bits of the identification of the reference subcarrier (IDID (RSID)) 21:20 2 Fixed two-bit identifier for one reference subcarrier Secondary Channel Pointer Binary Digit (CCI) 19 one 0 = primary only (hybrid or extended hybrid),
1 = primary and secondary (fully digital) Second binary parity eighteen one Parity check for IDVC, as well as zero and first IDPO Second Binary Timing Bit 17 one Second Binary Sync Bit = 0 First backup binary bit 16 one Managed by higher layers of the protocol stack from zero to third binary bits of the block count (SB (BC)) level L1 15:12 four A modulo 16 count that increments through 32 OFDM characters First binary parity eleven one Parity check for binary bits of SB from zero to third, as well as the first backup binary bit Zero and first binary bits of synchronization 10: 9 2 Zero and first binary bits of synchronization = 11 Binary digit of the pointer of the choice of the interleaver P3 (UVPR3) 8 one 0 = short length of interleaver P3 when using, 1 = long length of interleaver P3 when using Zero Standby Binary Bit 7 one Managed by higher layers of the protocol stack Binary digits (zero through fifth) of the Primary Service Mode Indicator (PSMI) 6: 1 6 Primary Maintenance Mode Value Zero Binary Parity 0 one Parity check for binary digits (from zero to fifth) of the primary service mode indicator, as well as for the binary discharge of UVPR3

Figure 10 presents a block diagram of a differential encoder 106. The binary bits in each column of the 32 × 61 matrix r collected by the system control data sequence assembly module are differential encoded in accordance with FIG. 10 and are output to obtain the matrix R in the same okay. Conceptually, this process can be considered as the work of parallel differential encoders, the number of which is 61. For a single differential encoder, the binary bits in one column j of the matrix r are processed sequentially, while i takes the values 0 ... 31. One binary bit is introduced into the differential encoder at a time system control data sequences. This input binary bit is subjected to summation modulo 2 with the previously stored output binary bit R [i-1] [j] to form the most recent output binary bit R [i] [j]. The resulting binary output bitstream will change polarity each time the input binary bit is set to 1. The initial state of each differential encoder is 0.

The reference subcarriers are modulated by a repeating 32-bit clock sequence of two-position (binary) phase shift keying (BPSK), which is subjected to differential encoding before transmission. Reference subcarriers serve several purposes: 1) to solve the problem of subcarrier ambiguity during capture, 2) to obtain a local phase reference signal for subsequent coherent detection, 3) local noise and / or interference samples to estimate channel state information (IOS) and 4) to obtain information phase error for tracking frequency and symbols. Differential coding of the clocking DPSK sequence allows the detection of the clocking DPSK sequence until the coherent reference signal necessary for the remaining subcarriers is established. The combination detected by differential coding is then used to remove modulation of the data from the reference subcarriers, leaving information about the local phase of the reference signal, as well as noise or interference samples. This is used to estimate the IOSK necessary for subsequent decoding in accordance with the soft decision.

11 shows a block diagram of a high level functional module of a coherent reference signal and IOSC. 11 binary digits of synchronization of a combination of synchronization of blocks is enough to uniquely identify the boundaries of each block, regardless of the values of the remaining binary digits, the number of which is 21. The DPSK for the reference subcarrier is chosen because this method is more tolerant to noise and channel deterioration than the four-position phase shift keying (QPSK) obtained by differential detection. In addition, the redundancy of the clocked DPSK sequence over all reference subcarriers gives a stable reference signal even under the most severe conditions of interference and channels. Various fields in the clocking DPSK sequence (hybrid or digital signal, backup, block counting, as well as mode) are checked for parity both to protect against errors and to prevent phase changes of the reference signal at the end of each changing field due to differential coding. The same clocking DPSK sequence (with the exception of the two-bit identifier field (ID) of the reference subcarrier) is excessively transmitted in all locations of the reference subcarriers and coincides with the interleaver block defined by the block counting field.

The coherent reference signal module and IOSC generates a phase reference signal for coherent detection of data subcarriers and appropriately weighs the resulting soft decisions based on the current state of the channel. Then, complex weighted soft solutions are scaled, limited, and stored in the appropriate matrices of the deinterleaver for output to the binary processing module.

The weighted soft decision module is smoothed over time to form a set of synchronization weights (weights) for the tracking (tracking) control module. In addition, soft solutions with full precision and floating point are also passed to the tracking control module to detect erroneous character tracking conditions.

On Fig presents a detailed block diagram of a high level functional module of the coherent reference signal and IOSK. At every moment of the presence of the OFDM symbol, the upper primary, lower primary and secondary symbols in the frequency domain subjected to phase alignment are received. Each symbol consists of a number of reference subcarriers and data subcarriers. Reference subcarriers are used to generate a coherent reference signal of the IOSK phase and weights, which are then applied to the data subcarriers to create soft solutions for processing binary bits and tracking control. In addition, IOSC weights are used to obtain synchronization weights for the tracking control module. The coherent reference signal module and IOSC are not executed until the frame synchronization and system control module detects the location of the border of the subframe (block of level L1).

The detailed block diagram shown in Fig. 12 shows that the module of the coherent reference signal and IOSc consists of seven main functional components:

a) component 110 to remove data from the reference subcarriers;

b) reference subcarrier estimation component 112;

c) a component 114 for estimating noise variance;

g) component 116 weighing the coherent reference phase signal;

d) a synchronization weight generation component 118;

e) reference subcarrier interpolation component 120;

g) soft decision management component 122.

The operation and interaction of these functional components is briefly described below. Using coded sequences of system control data from the frame synchronization / system control module, the data removal component from the reference subcarriers cuts data from each reference subcarrier subjected to phase alignment in the received OFDM symbol. Removing the modulation from the reference subcarriers provides an accurate estimate of the phase of the received signal for coherent detection.

The reference subcarriers, from which the data is cut, are then filtered - time and frequency - by the evaluation component of the reference subcarriers. The module of each filtered reference subcarrier serves to estimate its average signal power, and the angle provides the coherent phase reference signal necessary for the coherent detection of demodulated sub-carriers of the OFDM data.

In order to optimize the operation of the Viterbi decoder, the amplitude of each demodulated OFDM subcarrier must be weighted by evaluating its signal-to-noise ratio. This “IOSK estimate" is calculated and applied by weighting the coherent phase reference signal. The signal power estimate is simply a module of each filtered reference subcarrier, and the noise power estimate of each reference subcarrier is provided by the noise variance estimation component. The weighting component of the coherent phase reference signal divides the filtered reference subcarriers into their respective noise variance estimates, producing coherent phase reference signals weighted by IOSC for each reference subcarrier.

Then, the interpolation component of the reference subcarriers superimposes the coherent phase reference signals weighted by IOSK on the data subcarriers. This component interpolates between the coherent phase reference signals of the reference subcarriers and applies the interpolated reference signals to the corresponding data subcarriers. The resulting soft floating point solutions are passed to the tracking control module to detect erroneous character tracking conditions. In addition, the soft decision management component scales and quantizes soft floating point solutions, and also places them in the appropriate deinterleaver matrices for output to the signal processing modules and obtaining the bit error rate (BER).

The synchronization weight generation component computes a module of coherent phase reference signals weighted by IOSC, smooths them in time and passes them to the tracking control for use as synchronization weights.

Both primary and secondary sidebands are processed by the main functional components. The input signals to the coherent reference signal module and IOSC are aligned characters, encoded sequences of system control data, a group of segments, IOSK delay buffer information, synchronization status, and system control data.

Aligned symbols are used to generate a coherent reference signal of the phase and IOSK weights, which are applied to the data subcarriers to create soft output solutions. At each moment of the presence of an MORCH symbol, the upper primary aligned symbol and the lower primary aligned symbol are output from the phase alignment module. In addition, the phase equalization module also generates a secondary aligned symbol with a frequency R _s (344.53125 Hz) following the OFDM symbols. Each upper or lower primary symbol consists of 267 complex samples. Each secondary symbol consists of 559 complex samples.

The encoded sequences of system control data are used to cut data from the reference subcarriers in aligned characters. The frame synchronization / system control module supplies the primary and secondary encoded sequences of system control data to the coherent reference signal / IOS module. Four primary and four secondary sequences are supplied: one for each reference subcarrier identification value (IDPO). Each sequence provides a single value at the symbol rate.

Although all sidebands are processed by the same functional components in the coherent reference signal / IOC module, the subcarriers on which they operate can vary with the service mode. A group of segments identifies frequency segments that are filled within a given symbol. It indicates which primary extended segments and secondary sidebands are available for processing. At each symbol time, one value of a group of segments is supplied from the executive demodulator.

The IOSK delay buffer information is simply the address of the last element of the 22-character IOSK ring delay buffer. This buffer lines up the input aligned characters with their corresponding coherent phase reference signals weighted by IOSK. The IOSK delay buffer information is used to package the ring buffer pointers (links). There are two inputs of the IOSK delay buffer information: one for primary sidebands and one for secondary blockbands. Each input provides a single value at the symbol rate.

The synchronization state consists of two flags: one from the frame synchronization module (system control), indicating the actual block count, and the other from the executive demodulator, indicating that the input aligned symbols have undergone a sufficient delay to ensure proper alignment with their coherent phase reference signals weighted by IOSK. Both values of the synchronization state are provided at each symbol time.

The system control data supplied by the frame synchronization / system control module includes a primary maintenance mode, a secondary maintenance mode, and block counting. Service modes are used by the soft decision management component to determine the location and number of soft decision outputs. The block count indicates the number of the current L1 level block and is used by the soft decision control component to determine when soft decisions can be sent to the proper matrices of the deinterleaver. These three values of the system control data are provided at each symbol time.

The output signals of the coherent reference signal / IOS module are the matrices of the deinterleaver, the start flags of the deinterleaver, the primary soft solutions with complete accuracy and synchronization weight. At each symbol time, the coherent reference signal / IOC module sends quantized soft solutions to the primary and secondary matrices of the deinterleaver. The number of soft solutions and matrices of the deinterleaver for which they are intended are determined by the service mode. For each block of level L1, all soft decisions within the matrices of the inverted interleaver are passed to the binary processing modules and to obtain the FEC, primary soft decisions with full accuracy and synchronization weights by the executive demodulator.

The data deletion component of the reference subcarriers cuts the modulation from the input reference subcarriers subjected to phase alignment using sequences of encoded system control data from the system control frame synchronization module. This provides an accurate estimate of the phase of the received signal for coherent detection.

First, reference subcarriers are subtracted from input subcarriers subjected to phase alignment. The lower primary sideband contains reference subcarriers numbered 0 to 14, the secondary sidebands contain reference subcarriers numbered 15 to 45, and the upper primary sideband contains reference subcarriers numbered 46 to 60.

Modulation is removed from the reference subcarriers by multiplying the complex conjugate value of the reference subcarrier data by the complex data value of one of the four encoded sequences of system control data. The following equation illustrates this operation for each reference subcarrier within each OFDM symbol:

y = x ^∗ (1 + j) p _n ,

where y is the complex value of the cut out data of the reference subcarrier, x is the complex input value of the data of the reference subcarrier, n is the number (0, 1, 2, or 3) of the encoded sequence of system control data, and p _n is the proper value of the system control data:

The value of (1 + j) · p _n represents two points of the DPSK-set, not distorted by noise.

At each time of the OFDM symbol, one value is received from each sequence of system control data. The four encoded sequences of system control data received from the frame synchronization / system control module are modulated by four possible reference subcarrier identifier (IDPO) numbers. The data values of the reference subcarriers are multiplied by the values of the system control data that share the same IDIs.

The operation of deleting data from reference subcarriers for the upper primary sideband is illustrated in FIG. Data deletion operations from reference subcarriers for other sidebands are performed in a similar manner.

The reference subcarrier estimation component performs time and frequency filtering of the reference subcarriers from which the data is cut. The amplitude of each filtered reference subcarrier serves to estimate the average power of the signal, and the angle serves to estimate its coherent reference phase.

The reference subcarriers from which the data is cut are first passed through an 11-tap time domain filter with a finite impulse response (FIR filter). Since each reference subcarrier is filtered separately, there are essentially 15 parallel filters for both the upper and lower primary sidebands and 31 parallel filters for the secondary sidebands. The resulting reference subcarriers subjected to temporal filtering are then subjected to frequency filtering using a cascade of four 3-tap FIR filters. Each filtering stage is performed for all reference subcarriers within the OFDM symbol before moving on to the next stage. To fill the filter memory for external reference subcarriers, duplicate the end points (defined by the group of segments). Detailed flowcharts of a reference subcarrier estimation process for primary and secondary sidebands are respectively shown in FIGS. 14 and 15, respectively.

On Fig presents a block diagram of the assessment of the reference subcarriers of the primary side bands. On Fig presents a block diagram of the assessment of the reference subcarriers of the secondary side bands. The filtering operation in the time domain for each reference subcarrier in the OFDM symbol n has the following form:

where y [n] is the output reference subcarrier filtered in the time domain for the OFDM symbol n, x [n] is the input reference subcarrier subjected to cutting for the OFDM symbol n, h [k] is the array of FIR filter coefficients , and k is the summation index of the filter. For each OFDM symbol, all reference subcarriers are subjected to frequency filtering by solving the following equations in sequence:

where y [m] is the input reference subcarrier filtered in the time domain, h [k] is the array of FIR filter coefficients, and k is the filter summation index. In the above equations, m is the index of the reference subcarrier, the range of which for a particular sideband is a function of a group of segments. In addition, a [m], b [m], c [m] and d [m] are the output signals of the first, second, third and final stages of the FIR filter, respectively.

During the frequency filtering operation, the first and last reference subcarriers in the sideband are repeated at the beginning and end of each filtering step, respectively, to fill the filter memory for filtering the first and last subcarriers. For the upper primary sideband, the following reference subcarriers are duplicated as a function of a group of segments: the reference subcarrier at number 60 is duplicated and placed in an excess memory cell for all groups of segments; the reference subcarrier at number 50 is duplicated and placed in the position of the reference subcarrier at number 49 for the group of segments "Extension 0" (mode MP1 service); the reference subcarrier at number 49 is duplicated and placed in the position of the reference subcarrier at number 48 for the group of segments "Extension 1" (mode MP2 service); the reference subcarrier at number 48 is duplicated and placed in the position of the reference subcarrier at number 47 for the group of segments "Extension 2" (mode MP3 service); and the reference subcarrier at number 46 is duplicated and placed in the redundant memory cell for the group of segments "Extension 4" (modes MP4, MP5, MP6 and MP7 service). Filtering for other sidebands is done in a similar way.

The noise variance estimation component computes an estimate of the noise power for each reference subcarrier. This noise power estimate is used by the weighting component of the coherent phase reference signal to weight the filtered cut-out reference sub-carriers using the IOSK estimates.

The filtered reference subcarriers from the estimated component of the reference subcarriers (signal) are subtracted from their respective delayed reference subcarriers subjected to cutting (signal plus noise) to form integrated noise estimates (signal + (noise - signal) = noise) for each reference subcarrier in the primary and secondary sidebands. The squared module of complex noise estimates is then subjected to frequency and time filtering to obtain averaged estimates of noise variance for each reference subcarrier.

The operation of squaring the module is carried out at all noise values of the reference subcarriers, which leads to obtaining samples of the noise variance. Then, the noise variance estimates subjected to frequency filtering are passed into filters with an infinite impulse response (IIR filters) to estimate the noise. There are 15 separate IIR filters for both the upper and lower primary sidebands, as well as 31 separate IIR filters for the secondary sidebands (one filter for each sample of the noise dispersion of the reference subcarrier). IIR filters are lossy and unity gain bipolar integrators that average the input samples of the noise variance. Such operations are carried out one at a time at the presence of an OFDM symbol. An estimated average noise variance for each reference subcarrier is generated from these filters.

Detailed flowcharts of a noise variance estimation process for primary and secondary sidebands are shown in FIGS. 16 and 17, respectively. On Fig presents a block diagram of the estimation of the variance of the noise of the primary side bands. 17 is a flowchart for estimating the noise variance of the secondary sidebands.

The subtraction and squaring of the module for each reference subcarrier in the OFDM symbol n are described by the following equation:

z [n] = | x [n-5] -y [n] | ²

where z [n] is the noise variance sample, y [n] is the filtered reference subcarrier, and x [n-5] is the clipped reference subcarrier that is delayed by five characters to account for the FIR filter delay in the time domain in the reference estimate component subcarrier.

The frequency filtering operation is identical to that used in the evaluation component of the reference subcarriers. The FIR filtering operation for each reference subcarrier in the OFDM symbol has the following form:

- оценка выходной дисперсии шума для МОРЧ-символа n, а v[n] - оценка входной дисперсии шума, подвергнутой частотной фильтрации. Коэффициенты фильтра вычисляют на основании двухполюсного интегратора с потерями и с α=1/16:Where

is the estimate of the output noise variance for the OFDM symbol n, and v [n] is the estimate of the input noise variance subjected to frequency filtering. The filter coefficients are calculated based on the bipolar integrator with losses and with α = 1/16:

A = α ² = 0.00390625;

B = 2 · (1-α) = 1.875;

C = - (1-α) ² = -0.87890625.

The weight component of the coherent phase reference signal calculates and applies IOSK weights to each filtered reference subcarrier. These IOSK weights, which serve to estimate the signal-to-noise ratio of each reference subcarrier, are ultimately applied to each data subcarrier to improve the performance of the Viterbi decoder. Since the filtered reference subcarrier module displays the signal strength, the IOSK weight is calculated and applied simply by dividing each filtered reference subcarrier by its corresponding noise variance estimate. The module of the resulting complex reference subcarrier represents the weight of IOSC, and the angle is the coherent phase reference signal. The embodiment of this component is described below.

The filtered reference subcarriers are delayed by 16 characters to account for the IIR filter delay in the noise variance estimation component. To prevent division by zero, a small bias is first added to the noise variance estimates. Then, the delayed filtered reference subcarriers are weighted by dividing by the reciprocal of their corresponding noise variance estimates, resulting in coherent phase reference signals weighted by IOSC for each reference subcarrier in the primary and secondary sidebands.

On Fig presents a block diagram of the weighing IOSK reference subcarriers. The IOSK weighing operation for each reference subcarrier in the OFDM symbol has the following form:

- оценка входной дисперсии шума для МОРЧ-символа n, x[n-16] - входная задержанная отфильтрованная опорная поднесущая, а ε=10^-11 - смещение, добавленное для предотвращения деления на нуль. Операция взвешивания по ИоСК идентична для всех боковых полос.where w [n] is the output reference subcarrier weighted according to IOSK for the OFDM symbol n,

is the estimate of the input noise variance for the OFDM symbol n, x [n-16] is the input delayed filtered reference sub-carrier, and ε = 10 ^-11 is the offset added to prevent division by zero. The IOSK weighing operation is identical for all sidebands.

Sync weights are simply IOSK weights that are averaged over a long period of time to remove the effects of signal fading. The tracking control module uses synchronization weights to correct tracking for its respective symbol tracking loop.

Synchronization weights are generated for each OFDM symbol by calculating the modulus of each filtered subcarrier weighted by IOSK in the primary sidebands and smoothing the result using an IIR filter. There are a total of 30 separate IIR filters (one for each reference subcarrier). IIR filters are lossy and single gain bipolar integrators that average the IOSK weights over a long period of time. These operations are performed once at each time of the OFDM symbol.

All 30 synchronization weights are always generated - even in maintenance modes in which extended sidebands are not filled. This is acceptable because the IOSK weights (and therefore the synchronization weights) of the extended subcarriers are automatically minimized if they are not present.

On Fig presents a block diagram of the generation of weights synchronization. The complex module of the primary reference subcarrier weighted by IOSK in the OFDM symbol is calculated as follows:

where x _i and x _q are the real and imaginary parts of the input reference subcarrier, filtered and weighted according to IOSC, and w is the corresponding weight of IOSC.

The IIR filtering operation for each primary reference subcarrier of the OFDM symbol n has the following form:

y [n] = A · w [n] + B · y [n-1] + C · y [n-2],

y [n] is the output synchronization weight associated with a particular reference subcarrier in the OFDM symbol n, and w [n] is the input weight of IOSC. The filter coefficients are calculated based on the bipolar integrator with losses and with α = 1/128:

A = α ² = 0.00006103515625;

B = 2 · (1-α) = 1.984375;

C = - (1-α) ² = -0.984436035.

The operation of generating synchronization weights is carried out only on the primary sidebands.

The reference subcarrier interpolation component applies coherent phase reference signals weighted by IOSC to each data subcarrier. A coherent phase reference signal is generated for each data subcarrier by interpolation between the coherent phase reference signals of the reference subcarriers. The resulting coherent phase signals of the data subcarriers are then multiplied by the corresponding data subcarriers to obtain coherently detectable, IOSC-weighted complex soft decisions.

For primary sidebands, the filtered reference subcarriers weighted by IOSC are passed through a frequency domain filter that has an interpolation coefficient of 19. This interpolation filter is a 37-tap FIR filter that “slides” over all active reference subcarriers. Since it does not perform temporary filtering, the interpolation filter only works on data from one OFDM character.

In the process of interpolation, 18 complex zeros are first inserted in the spaces between each filtered reference, which realizes the interpolation coefficient 19. In addition, 18 zeros are added before the first reference subcarrier and after the last reference subcarrier - for proper filter initialization. After that, the FIR filter works on data filled with zeros. When it “glides” in frequency, there will be at least two reference signals (non-zero input signals) at some point in time within the filter. When the filter is centered on the reference signal, there is only one reference signal within the filter, and it is aligned with the center tap. In this case, the filter simply returns the input reference subcarrier, which is later discarded.

FIG. 20 is a schematic illustration of an illustration of interpolation of reference subcarriers for primary sidebands. Interpolation filtering for the secondary sidebands is carried out in a similar manner.

Immediately after a certain symbol is filtered out, the reference subcarriers are discarded from the output signal, and only coherent phase reference signals, weighted by IOSK, remain for each data subcarrier. The interpolated phase reference signals are then multiplied by their aligned data subcarriers to create soft output solutions. Before complex multiplication, the data subcarriers subjected to phase alignment must be properly synchronized with the interpolated phase reference signals. This is achieved by delaying the data subcarriers subjected to phase alignment by 21 symbols. The delay takes into account the total delay introduced by the FIR filter, filtering the reference subcarriers in the time domain (5 characters), and the IIR filter of the noise variance estimate (16 characters).

On Fig presents a block diagram of the interpolation of the reference subcarriers. The interpolation filtering operation for each subcarrier n in the OFDM symbol has the following form:

- заполненный нулями входной сигнал для МОРЧ-поднесущей

, y[n] - интерполированный когерентный опорный сигнал фазы для МОРЧ-поднесущей n, h[k] - массив коэффициентов КИХ-фильтра, а k - индекс суммирования фильтра. Количество К отводов фильтра составляет 25 для вторичных защищенных боковых полос и 37 - для всех остальных боковых полос.Where

- a zero-filled input signal for the OFDM subcarrier

, y [n] is the interpolated coherent phase reference signal for the OFDM subcarrier n, h [k] is the array of FIR filter coefficients, and k is the filter summation index. The number of filter taps K is 25 for the secondary protected sidebands and 37 for all other sidebands.

To minimize channel losses, Viterbi decoding with weighting for summing maximum ratios (MRC) for coherently detected MPSK subcarrier symbols is used for soft solutions. Since the levels of interference and signals change with the change of subcarriers (frequency) and time due to selective fading, over time IOSK becomes necessary for adaptive weighting for soft decision symbols used as branch metrics for Viterbi decoding. The IOSK estimation method should be developed taking into account the fading bandwidth of up to about 13 Hz for maximum vehicle speeds in the FM broadcast band of about 100 MHz. A delay spread of several microseconds is typical, although judging by the measurement results, there were even more variations in some environments. A method for evaluating both the phase reference signal and the IOSK based on the reference subcarriers is illustrated in FIG. Obtaining this IOSK weight involves weighting the amplitudes for the QS along with phase correction for phase errors in the channel.

Operation in accordance with the IOSK recovery method of FIG. 21 involves frequency capture and tracking of subcarrier frequencies, as well as time matching of symbols with respect to OFDM symbols. Methods of capturing frequency and clocking of symbols involve the use of cyclic prefix properties. Frequency and symbol tracking is achieved by observing the phase drift from symbol to symbol depending on time or frequency (among subcarriers).

After capturing the frequency and matching the symbols in time, an attempt is made to synchronize the combination of synchronization blocks of the clocking DPSK sequence by cross-correlating the differential detected DPSK sequence with the combination of block synchronization (Block Sync). Differential detection is performed on all subcarriers under the assumption that the location of the “training” subcarriers is unknown at first. Cross-correlation of a known combination of block synchronization with detected bits of each subcarrier is performed. Subcarrier correlation is recognized when a match of all 11 bits of a block synchronization combination is detected. Block synchronization (and the resolution of the subcarrier ambiguity problem) is established when the number of subcarrier correlations matches or exceeds threshold criteria (for example, 4 subcarrier correlations spaced into several subcarriers, the number of which is a multiple of 19 subcarriers).

After the synchronization of the blocks is established, the changing fields of the clocked DPSK sequence can be decoded. Differentially detectable binary bits of these variable fields are detected based on a “majority vote” of the “training” subcarriers, so decoding is possible when some of these subcarriers or binary bits are distorted. 16 blocks within each frame of the modem (Modem Frame) are given numbers from 0 to 15. Then the contents of the most significant binary digit (SSB (MSB)) of the Block Count field are always set to zero, since the block count never exceeds 15. The modem frame synchronization is set in accordance with the knowledge of the “block counting” field.

Coherent detection of this signal requires some coherent phase reference signal. The decoded information from the clocked DPSK sequence is used to remove modulation from the “training” subcarriers, leaving information about the reference signal of the local phase and noise. Turning to FIG. 26, we note that first, from the received “training” subcarriers, the modulation of the binary (± 1) clock sequence is removed, multiplying the mentioned subcarriers by a synchronized, decoded, and differential transcoded clock DPSK sequence. To smooth the resulting symbols in time, use an FIR filter, which gives a complex conjugate estimate of the local phase and amplitude. This value is delayed and multiplied by an estimate of the reciprocal of the noise variance. The noise variance is estimated by subtracting the smoothed estimate of the local phase and amplitude from the input symbols (after proper synchronization), by squaring and filtering the complex noise samples, followed by approximation of the inverse value (with protection against division by zero). This weight IOC is interpolated over 18 subcarriers between pairs of neighboring “training” subcarriers. The resulting local IOSK weights are then used to multiply by the corresponding local characters carrying the data.

To evaluate the complex coherent reference signal α at each subcarrier location, an 11-tap FIR filter is used. A larger FIR filter with more taps could reduce the estimation error when the signal statistics are constant, but the bandwidth would be too small to track Doppler-like changes in the signal at the maximum speeds typical of freeways. Therefore, 11-tap filters with a wedge-shaped symmetrical Gaussian impulse response are preferred. Instead of the IIR filter, a symmetric FIR filter is used because of its linear phase property, which gives a zero bias error for the (approximately) piecewise linear fading characteristic in the channel in the filter frequency range. This smoothed coherent reference signal output from the FIR filter is subtracted from the delayed input samples, which gives instant noise samples. These noise samples are squared and processed by the IIR filter, which gives an estimate of the noise variance, σ ² . This filter has a bandwidth that is narrower than the FIR filter, which allows a generally more accurate estimate of the noise variance. After appropriate sample delays, to match the filter delays, a symbol weight α ^∗ / σ ^{2 is} calculated for each subcarrier. These values are smoothed and interpolated by subcarriers for each OFDM symbol to obtain more accurate estimates. This weight is unique for each OFDM symbol, and each subcarrier provides a local (time and frequency) estimate and weight for the symbols forming the branch metrics for the subsequent Viterbi decoder.

The system just described was designed to match vehicles with fixed antennas. Numerous functions of reference subcarriers are described in relation to capture, tracking, estimation of channel state information (IOSC) and coherent operation. This system was designed for coherent operation in the FM (FM) broadcast band (88-108 MHz) with a fade bandwidth corresponding to vehicles traveling at speeds characteristic of freeways. Various coherent tracking parameters are estimated using filters whose bandwidths approach the expected width (approximately 13 Hz) of the Doppler frequency band. With a stationary antenna, the corresponding tracking statistics for the input signal according to tracking algorithms are assumed to vary with a frequency not exceeding the Doppler frequency bandwidth. However, the use of patch diversity antennas in vehicle windows introduces abrupt transients into the coherent tracking of a digital signal, which degrades digital performance. An example of such an antenna diversity switching system is shown in US Pat. No. 6,633,258.

The antenna diversity switching system includes several antenna elements (for example, from 2 to 4), which are usually located inside the glass of the front or rear windows of the vehicle. These elements are connected to a diversity switching module that dynamically selects one element or combination of elements to provide an RF antenna signal to the receiver. A functional diagram of a diversity switching configuration is shown in FIG. Several elements — 200, 202, 204, and 206 — are connected to the diversity switching module 208. The diversity switching module includes an input 210 for receiving a control signal. The control signal is compared with the evaluation threshold, as shown in block 212, and the control signal and the comparison results are input to the amplifier 214. The switching control unit 216 responds to the output of the amplifier by controlling the antenna switch 218. This switch connects one of the antennas to the antenna input 220 of the receiver.

The diversity switching module monitors the signal from the receiver to determine when to switch. A typical modular "blind switching" algorithm sets the switching threshold based on the average level of the intermediate frequency (IF) signal of the receiver. When the IF signal falls below this threshold, the switch blindly selects an alternate item in anticipation of a better signal. If the new signal exceeds the threshold, then the switch supports the selection of a new element; otherwise, the diversity switching module selects an alternative element after a certain minimum response delay time has elapsed. This process continues, and the diversity switching module continuously updates its threshold (its thresholds).

On Fig presents a functional diagram of a characteristic antenna system with switching diversity. The theory underlying the work of the diversity switching algorithm is based on different instantaneous fading characteristics of various antenna elements. Multipath fading causes an increase in the multiple beams (multipath) of the signal arriving at the receiving antenna element at different points in time. For example, a wavelength at 100 MHz is approximately 10 feet. If two signal propagations arrive at time points differing by one wavelength, i.e. 10 nanoseconds (10 ft. Travel difference), these signals will add up in phase. Similarly, if two beams arrive at the antenna element with a time difference corresponding to the half-length of the wave, then these signals folding in antiphase will be mutually destroyed. This addition or mutual annihilation is dynamic in a moving vehicle, where the Doppler frequency bandwidth is approximately determined by the expression BW = f _c · s / c (where f _c is the carrier frequency, s is the vehicle speed, c is the speed of light). The Doppler frequency bandwidth is approximately 10 Hz at speeds typical of freeways. Therefore, the signal vector (a complex version of the amplitude and phase) of one antenna element can change with a frequency of about 10 Hz in this example. Then the coherent tracking of the reference signal and the channel state must be consistent with a bandwidth of 10 Hz in order to maintain coherent tracking of the signal.

Typical antenna elements in a vehicle may be subject to some instantaneous fading conditions (depending on the gap between the elements and the directions of multipath propagation). For example, one element may be at zero fading, and another element at maximum. In a vehicle with several such elements, it is likely that some alternative element will receive a significantly larger signal when this element experiences fading (mutual annihilation of signals). Typical elements in a multi-diversity FM antenna system with diversity will have instantaneous fading characteristics that could be made correlated to some degree, but they are left uncorrelated enough to achieve the desired diversity gain to improve performance. On Fig presents a graph of the actual 226 and imaginary 228 components of the signal fading (no noise) in the case of an antenna element, switched every 100 characters.

In this example, it is assumed that the switching time characteristic of the diversity switching module is of the order of 10 microseconds, and the minimum response delay time is of the order of 10 milliseconds. The switching time of the order of 10 microseconds does not have a noticeable effect on the demodulated FM signal, and the overall improvement in diversity with frequency modulation can be significant, since signal fading is prevented. However, the switching transient negatively affects a coherently monitored digital signal. The challenge is to find a coherent tracking method that compensates for switching transients in an antenna diversity switching system.

The problem can be summarized as follows. Switching the antenna module causes a jumplike transient in a coherent signal, and the signal itself immediately before or after the transient is approximately coherent. Filtering with a finite impulse response (FIR filtering) of the hopping transient distorts the estimated reference signal α of the coherent channel, affecting the range of samples equal to the length of the FIR filter (11 taps). This distortion in the reference signal of the coherent channel increases the modulus of the noise samples by subtracting this reference signal from the input samples. Although noise increases only in the vicinity of the transient, a filter with an infinite impulse response (IIR filter) extends this noise peak to the range of the IIR filter. This gives the effect of underestimating the estimate of the noise variance in the immediate vicinity of the transient, while an overestimation of the IIR filter in the time range in the immediate vicinity of the transient does not occur. Both the distorted reference signal of the coherent channel and the noise dispersion errors contribute to the degradation of the digital signal.

On Fig presents a graph of the real and imaginary components of the filtered coherent reference fading signal in the vicinity of the transient on the symbol at number 200 with a signal-to-noise ratio (SNR) of 10 dB. 24 illustrates an estimated coherent reference signal in the vicinity of an intermittent switching transient. The solid lines 230 and 232 represent the smoothed values (α) of the complex channel gain and show the effect of the 11-tap FIR filter, in which the reference signal is distorted in the vicinity of the transient. The FIR filter has the effect of smoothing interpolation (distortion) of samples from either side of the switching process. The dashed lines 234 and 236 illustrate a better estimate of the coherent signal, in which only the symbol affected by the transient is affected.

25 is a graph illustrating the effects of switching transients on noise estimates (SNR 20 dB). Line 240 displays the noise sample input to the IIR filter, and line 242 displays the filtered output. It can be shown that the input noise samples (squared) increase in the immediate vicinity of the transient. In fact, this is due to the distortion of the reference signal of the coherent channel due to the 11-tap FIR filter. Although the output of the IIR filter responds to noise peaks, these peaks undergo local suppression at the transient, but propagate throughout the IIR filter. These errors in local estimation of noise variance contribute to poor digital performance.

The solution can be summarized as follows. Coherent tracking can be achieved on either side (in time) of the switching transient, as shown previously in FIG. 24. Coherent estimation filters cannot use samples of signals that span the transient on both sides, so tracking filters can only use signal samples up to the transient (not inclusive) from any direction in time.

One possible way to correct the error of a stepwise transient is to identify the temporary location of the transient. Although (theoretically) there should be the possibility of implicit communication of the switching moment to the receiver’s modem, this may turn out to be practically impossible. For the modem, reliable transient detection within the framework of coherent tracking algorithms is preferred. Coherent tracking algorithms can be modified in the vicinity of the transient to provide approximately coherent tracking in the immediate vicinity of the transient, with the exception of the symbol where the transient occurred. Needless to say, an error in the estimation of noise variance arises from an error in the reference signal of the coherent channel.

The complexity of the transient detection algorithm and the deterioration caused by the false transient detection stimulated the development of a much simpler algorithm, which led to improved performance.

In the channel of averaged white Gaussian noise (AWGN), linear filtering of the reference signal of the coherent channel and estimation of the noise yield almost optimal results. The type of linear filter and the time range of the filter depend on the fading statistics, and more specifically on the autocorrelation function of the parameter of interest. A symmetric 11-branch FIR filter approximating this autocorrelation function at speeds characteristic of freeways was used in the original design as a reasonable compromise between reducing the estimation error (dispersion) and the estimation speed for tracking the parameter during fading. An IIR filter was used to estimate the noise variance, since the linear phase does not cause problems in the presence of noise. Both of these estimation filters have been enhanced by additional frequency filtering by subcarriers using an effective frequency range that satisfies the selective fading environment (frequency autocorrelation). These filters give good results in the UBGS channel, with fading and assuming a reasonably slow automatic gain control (AGC) function, which should be slower than the fading dynamics. On Fig presents a block diagram of a coherent estimate of the reference signal and noise.

Some automotive receivers employ diversity diversity antennas and faster response AGCs. Coherent reference and noise filter filters are no longer suitable for these applications. In addition, the interference from the first neighboring FM signals is not UBGS and has a more impulse-like characteristic, as a result of which some improvement can be achieved by appropriate modifications to the methods for implementing evaluation filters. Transient detection and subsequent adjustment of the coherent reference signal can be used along with a modification of the noise estimate to solve this problem.

It is well known that median filtering - instead of linear (FIR- or IIR-) filtering - is a way to reduce estimation noise while maintaining spasmodic changes in the estimation parameters, as well as reducing the effect of impulse noise. This median filtering can conveniently replace transient detection and regulation of a coherent reference signal. Some important characteristics of the median filter are analyzed and described below to demonstrate their suitability for use in the receiving system.

One important characteristic of a median filter is to reduce the variance of the error estimate of the coherent reference signal and noise. This decrease is comparable to the original filters, and also comparable to the Cramer-Rao limit, which is the smallest possible estimation error for any filter (linear or non-linear) designed for a given number of samples (i.e., having a certain number of filter taps). In this case, the estimation error is superimposed only on such noise as UBSH.

First, we analyze the original 11-tap FIR filter used for coherent estimation of the reference signal. The 11 filter taps are normalized so that the sum of the corresponding coefficients is equal to unity, which gives a constant current gain of unity. On Fig presents a graph in the case of an 11-tap FIR filter for coherent estimation of the reference signal of a single reference subcarrier.

The following describes the filter used to smooth the estimates of the reference subcarriers. This filter has 9 taps (active) normalized in such a way that the sum of the corresponding coefficients is equal to unity, which gives a constant current gain of unity. The filtering of subcarriers at the ends of the upper or lower digital sidebands is not accurate due to truncation of the filter range at the end points. This filter is described in FIG. 28 illustrates a graph in the case of using a frequency filter to obtain reference subcarriers.

. Уменьшение дисперсии 11-отводного фильтра составляет 0,127 или около -9 дБ. Аналогичное уменьшение для 9-отводного частотного КИХ-фильтра составляет 0,196 или около -7 дБ. Рабочие характеристики уменьшения ошибки этих фильтров можно сравнить с медианными фильтрами и пределом Крамера-Рао.The following is a comparison of the reduction of the estimation error. The ratio of the variance of the output samples of the filter to the variance of the input samples — assuming that some constant parameter is distorted by the UBGS — is equal to the sum of the squares of the normalized filter coefficients, decrease_disp =

. The dispersion reduction of the 11-tap filter is 0.127 or about -9 dB. A similar decrease for the 9-tap FIRC filter is 0.196 or about -7 dB. The performance of the error reduction of these filters can be compared with the median filters and the Cramer-Rao limit.

For samples of a constant (over a filter range with N taps) parameter α distorted by the UBGS, the logarithmic likelihood function has the form:

and then the resulting Cramer-Rao limit for a constant parameter under the conditions of UBSH is simplified, taking the form:

Then the theoretical minimum dispersion for a filter with N taps is simply the input dispersion divided by the number of taps. This is consistent with the assumption that with equal (identical) branch weights, a minimum dispersion estimate cannot be achieved; however, this would work well when fading. On Fig presents a graph of the effects of the type of filter to reduce the estimation error for a coherent reference signal.

Although the median filter gives an increase in error dispersion of about 2 dB relative to the Cramer-Rao limit under UBSH conditions, real bell-shaped FIR filters allow for performance characteristics closer to median filters with the same number of taps. For noise distributions that have slowly decreasing “tails” or are more impulse-like than Gaussian noise, the performance of the median filter increases, while the performance of the linear filter remains the same. As an example with Laplace noise, we note that the working characteristic of the median filter is 3 dB better with a decrease in the variance of the error than a linear filter with the same number of taps. Of course, the Cramer-Rao limit for this new case no longer satisfies the FIR filter with the same tap weights.

The performance degradation under UBSH due to these filters can be approximated by calculating the additional noise (variance of the estimation error), in addition to the variance of the actual noise of the characters, during the detection of each character. The 9-tap filter of the frequency range is used independently of the 11-tap FIR or median filter performing temporary filtering. The 9-tap frequency filter does not affect the response to the jump (time) due to switching. Losses for the cases of the initial 11-tap FIR filter and 7-tap and 5-tap median filters are summarized in Table 2.

table 2
Filter loss Filter type 11 bypass FIR 5-tap median 7-tap median Error variance 0.127 0.314 0.224 With frequency filtering via a 9-way filter (0.196) 0,026 0,062 0,044 Losses in the conditions of UBSH 0.11 dB 0.26 dB 0.19 dB

On Fig presents a graph comparing the 11-tap FIR filtering and median filtering conducted on a coherent reference signal. Although a small loss of AWGN (UBGS) performance is expected when using the median filter, the improved response to the jump during antenna switching clearly follows from the graph shown in FIG. Note that the response to the jump is significantly improved without requiring a transient detection. A functional diagram of the new reference signal and noise estimate processing is shown in FIG. 31, and here the 11-tap FIR filter is replaced by a 5-tap median filter.

The following is a brief description of the functions and algorithms of FIG. The purpose of the process (s) shown in this drawing is to provide estimates of the complex gain (values “a”) of the coherent channel along with estimates of noise or interference. These estimates are local in time and frequency (related to the location of the carrier) and are designed to account for the channel with dynamic selective fading existing in the environment of a moving object, such as a moving car. These estimates are derived based on the symbols of the reference subcarriers, which are cut from the received and demodulated signal, as described above, and entered over the bus 250 as S _{r, n} complex values. The data used to modulate these symbols is already known and removed from these symbols through the first multiplication by the conjugate (illustrated by the multiplication block 252) to obtain the instantaneous values a2 _{r, n of the} complex channel gain on the bus 254. The subsequent median time filtering in filter 256 reduces noise while maintaining spasmodic changes due to antenna switching to obtain intermediate values a1 _{r, n} on bus 258. These intermediate values are then exposed (frequency) filtering (smoothing) by reference subcarriers, as shown in block 260, to obtain the final values of a _{r, n of the} complex channel gain. These gain values a _{r, n are} subsequently used outside this algorithm for processing (correction and providing information about transition metrics) as applied to signal sets for symbols carrying data, in the usual way when modulating symbols by the method of quadrature amplitude modulation (QAM).

The next step in this process is to evaluate the noise associated with each of these complex channel gain values. Instant noise samples are estimated by subtracting the values of a _{r, n-2} from the (appropriately delayed) noisy corresponding input samples a2 _{r, n-2} , as illustrated at summation point 262. As shown in block 264, based on these complex noise samples, squared values are computed to obtain instantaneous estimates of “dispersion _n-2 ” noise variance on bus 266. These instantaneous noise variance samples are poor estimates of local (temporal and frequency) noise and require processing and filtering to obtain useful estimates of noise variance. Although simpler time and frequency filtering can usually be used to reduce the error of these instantaneous estimates of noise variance, this type of filtering would be ineffective at taking into account the changing noise due to fading, AGC effects and spasmodic changes due to antenna switching. Therefore, when the temporal filter these instantaneous variance samples using a median filter 268 to obtain samples varflt _{n-16 (VAR-n-filter 16),} and the usual filtration (through a linear IIR or FIR filter 270) is used to further smoothing in frequency ( by subcarriers) in order to obtain final estimates of σ ² _{r, n-16 in a} manner analogous to the above-described way of obtaining estimates of the complex channel gain. To capture the relatively large noise pulses that arise due to the switching of the antenna, an additional direct transmission path 272 is provided. When these values (scaled by multiplying by a factor of 0.5, as shown in block 274) exceed the estimate obtained by median filtering, these higher values are selected to be output to the frequency smoothing filter by the maximum selection function illustrated in block 276. Then these the values are smoothed by reference subcarriers, as shown in block 278. This will be important in the subsequent formation of transition metrics, which involves the use of this knowledge of large noise pulses.

The just described analyzes and modeling of the improvements in the algorithms introduced into the coherent estimation of the reference signal obviously work quite well in all analyzed and simulated cases. These cases include a channel with smooth and selective fading and with a Doppler frequency bandwidth corresponding to the speeds typical of freeways and such insignificant noise as the corresponding 0 dB SNR. At the same time, other conditions in the channel should be considered, such as impulse noise or residual effects of transients, not completely suppressed by the new coherent processing of the reference signal. In this case, the adjustable x values of the coherent reference signal are suitable; however, the estimate of noise variance may be biased. In the case of a symbol (s) in which the impulse is distorted, the noise impulse may be high, but the IIR filter could suppress this value of the noise estimate at the moment of appearance of the impulse and extend the noise estimate over the reaction time of the IIR filter to the impulse disturbance. In this case, it would be preferable to directly transmit the loud noise samples parallel to the IIR filtering path (with suitable delay matching). For characters in which the noise pulse is significantly larger than the output of the IIR filter, this noise pulse should be used to determine the estimated noise variance for these symbols. When using the direct transmission path for these noise pulses, the energy entering the IIR filter for these samples should be reduced so that the local noise peak does not propagate over the IIR filter range. It is easy to envision several variations of this process for processing noise peaks in the estimation of noise variance.

The noise variance estimation process is modified to improve performance using switching processes and to align with faster AGCs. The initial noise estimate used a 2-pole IIR filter with parameter a = 1/16 (not to be confused with the described designations “a _{r, n} ” values for complex channel gains). The peak of the impulse response of this filter occurred with a delay of 8 samples (symbols), although the decaying “tail” was much longer, making the jump delay closer to 16 samples (symbols). Fig. 32 shows that the new modification changes the parameter “a” of the IIR filter, as a result of which a = 1/8. Other modifications to the noise estimation include changing the location of the frequency filtering, i.e. holding it not before the IIR filter, but after the IIR filter. The IIR filter is also preceded by a 7-tap median filter, and an excess direct noise path is provided. Both of these innovations are designed to better match transient switching processes and pulse-shaped noise, while matching with the faster AGC function in the input stage or inverter IF section.

A 7-tap median filter is used to pre-process samples of effective noise to a linear IIR filter and for frequency filtering (reference subcarriers). In this case, the median filter produces a biased estimate of the variance due to the property of the asymmetric probability density function (FPF) of squared noise samples. This offset can be adjusted for subsequent work in the noise estimation process. In particular, each input sample x entering the median filter is assumed to consist of the sum of a pair of (complex) squares of Gaussian noise samples with zero mean, where x = v ² + w ² . Gaussian FPV noise samples u and v have the form:

The cumulative distribution function of x can be found by changing the variables and integrating as follows:

Then find the probability density function (fpv) from x, differentiating P (x).

The noise variance is defined as the average value of x.

The median value of x can be found by solving the following equation for medx (median value of x):

The bias coefficient of the median estimate is found as the ratio of the median value to the average value.

Therefore, the median value causes an underestimation of the estimation of the dispersion of Gaussian noise with coefficient ln (2). This bias should be considered when it is used to replace the average estimate in the noise variance estimation process.

The scaling of the noise estimate due to the bias factor of the median value should usually not matter if the noise estimates for all the symbols presented in the Viterbi decoder are scaled with the same coefficient. In addition, nonlinearity can also have an effect in the formation of transition metrics (for example, in the case of a fixed quantization and overflow point).

The effects of decreasing the IIR filter time constant during dispersion estimation are discussed below. Each output sample y _{n of a} bipolar IIR filter is a function of a new input sample x _n and two previous output samples.

y _n = a ² · x _n + 2 · (1-a) · y _n-1 - (1-a) ² · y _n-2 .

The filter expression has a constant current gain of unity equal to unity, and its impulse response (in time) depends on the parameter “a”. The peak of the impulse response (mode) occurs at half the group delay time, while the group delay is more indicative of the response time to the jump. The mean and mode delays can be calculated as a function of parameter “a”. The impulse response and the response to the IIR filter jump for a = 1/16 and a = 1/8 is shown in Figs. 32 and 33. Fig. 32 is a graph of the impulse response of the IIR filter. On Fig presents a graph of the response to the jump of the IIR filter.

The effective filter delay should be equal to at least the mode delay, but less than the group delay. This delay should be adequately compensated when noise estimates are applied in the formation of branch transition metrics.

group delay = average delay = 2 · (1 / a-1),

fashion_delay = 1 / a-1.

The ability to suppress Gaussian noise can be estimated in the same way as in the case of an IIR filter, where the decrease in the variance of the noise estimate from input to output is equal to the sum of the squares of the normalized filter coefficients (DC gain is 1). Note that this decrease in dispersion is associated with the variance of the error in estimating the noise variance, and not with a decrease in the variance of the noise power. For an IIR filter, it is more convenient to carry out this calculation on the z-transform of the impulse response of the filter, assuming for large N the limit that serves as an approximation of the continuous spectrum, followed by the application of the Parseval theorem. In this case, the z-transformation for the IIR filter has the form:

and its module has the form:

The dispersion reduction coefficient from input to output samples can be calculated as follows:

Passing to the limit, we can integrate the sum as follows:

Then, the dispersion reduction coefficient is found as a function of parameter “a” after integration

The dispersion reduction achieved by the IIR filter with UBSH is plotted as a function of parameter “a” in FIG. 34. On Fig presents a graph of the reduction of errors in the estimation of noise as a function of the parameter "a" IIR filter.

Frequency filtering provides an additional 7 dB of noise reduction for UBSH. In the stationary UBSH channel, an increased IIR filter time constant would be beneficial to reduce noise estimation error; however, in a non-stationary channel with switching transients and / or with a faster AGC, a reduced time constant would be preferable. For the system in question, a 7-tap median filter followed by an IIR filter having a = 1/8 is preferable to the original design with only an IIR filter having a = 1/16. This new filtering leads to minimal degradation in the fixed channel of the UBGS, while providing significantly improved performance during fading with fast AGC and in antenna diversity switching processes. Then, the resulting error in noise estimation due to only the IIR filter and the frequency filter is approximately 22 dB better than the variance if a single noise signal is input to the filter. This leads to an error in estimating the noise variance, which has approximately 8% standard deviation of the noise variance at the input of the filter, or, equivalently, averaged over 150 noise samples. This can have an effect similar to the degradation of the transition metrics used by the Viterbi decoder by approximately 0.33 dB in UBSH. A cascaded 7-tap median filter reduces this error even more, although its effect in the UBGS is not easy to calculate. This decrease is comparable to a degradation of about 0.16 dB in the original modem, where a = 1/16 is used. At the same time, there is motivation to create a new noise estimate for conditions in which noise is different from UBSH.

In a channel where there is only UBSH, the optimal receiver may not select the redundant direct noise transmission path, and the optimal noise filter may not be linear (for example, an IIR or FIR filter), and its operation may take a long period of time. An excess noise path is provided only for processing pulse-shaped noise or coherent reference signal errors due to transient switching processes, etc. In these cases, an excess noise path mitigates the random effects of large noise bursts without spoiling the longer-term dispersion estimates provided by the IIR filter. The excess direct noise transmission path is scaled with the coefficient G ∗ ln (2), and the proportion of input noise samples in the noise estimation path can exceed the average value and can be selected using the “Select Maximum Input Signal” function. If the scaling is small, then the excess noise path may not be chosen so often, and losses in the case of UBSH can be minimized. The probability that a redundant path is selected is approximately equal to the probability that the redundant samples exceed the average value (assuming a filter that works exactly according to the average value).

The average increase in the variance of the noise estimate (in dB) due to the choice of the excess path can be calculated as follows:

Fig. 35 is a graph illustrating the effects of an excess gain value G ∗ ln (2) on the probability of selecting an excess path and on the increased average estimated noise variance. The value G ∗ ln (2) = 0.5 was empirically determined as a reasonable compromise value for the excess noise gain. This value gives minimal losses in the case of UBSH, but is effective in minimizing losses due to transient switching processes. In addition, performance is improved in the presence of fast AGC and / or first adjacent FM interference.

The effects of changes in both residual errors of the coherent estimate and noise variance estimates are shown in FIGS. 36-40. Fig. 36 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) in evaluating the original 11-tap FIR filter and the original IIR filter with a = 1/16. Fig. 37 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) when evaluating the initial 11-tap FIR filter and the IIR filter with a change according to which a = 1/8. Fig. 38 is a graph illustrating the effects of switching transients on noise estimates (10 dB SNR) in the presence of an initial 11-tap FIR filter when a variation in noise variance is superimposed. Fig. 39 is a graph illustrating the effects of switching transients on noise estimates (SNR 10 dB) in evaluating a 5-tap median filter and an IIR filter with a change according to which a = 1/8. 40 is a graph illustrating the effects of switching transients on noise estimates (SNR 10 dB) in the presence of a 5-tap median filter when a variation in noise variance is superimposed.

To obtain the graphs according to Figs. 37-40, various components of the changes are introduced, with Fig. 40 being the final result of all the changes. On Fig shows the operating characteristics of the source modem in terms of errors of coherent estimates of the reference signal and noise. This drawing shows the relatively high value of the transient noise during switching, which is not captured in the noise estimate. Noise peaks are underestimated in the estimation, and estimates in the vicinity of the peaks are overestimated due to smoothing by the IIR filter.

On Fig shows a slight improvement achieved by reducing the time constant of the IIR filter. FIG. 38 shows a significant improvement in noise estimation, although the actual transient noise remains large. On Fig shows a significant improvement in transient noise of the coherent reference signal through the use of a 5-tap median filter, although the noise estimate does not capture the transient peaks.

On Fig shows all the improvements that are the results of reducing the noise of the coherent reference signal, while the noise estimate captures the short duration of the transient processes. Although the noise variance estimates on the graph may seem somewhat noisy, this choice of parameters gave the best compromise in performance over the area of test scenarios.

This invention provides coordination of switching transients that an antenna diversity switching system has to deal with for hybrid FM signals of the HD system

^RadioTM These improvements entail changes in coherent tracking and noise variance estimation algorithms. New algorithms were analyzed and modeled with smooth and selective fading at speeds typical of freeways and Rayleigh fading. The simulation shows good performance of coherent tracking and an acceptable noise estimate. The improvements mentioned gave significantly better performance with fast AGC, antenna diversity switching, first adjacent FM interference, and in extreme selectable fading environments.

Laboratory tests of new software for the receiver of the reference signal (modified to control the antenna diversity switching) were conducted. The final option seems to provide acceptable operating gain factors in some key fading and interference environments. Fig. 41 quantitatively indicates the performance when fading at its natural frequency of oscillation (SCH), and Fig. 42 shows the performance when fading at ultrasonic frequencies (UHF).

FIG. 43 is a block diagram of a receiver 314 configured to process signals in accordance with this invention. HD signal

Radio ^{TM is} received at antenna 316. A pre-selection bandpass filter 318 passes the frequency band of interest, including the desired signal at f _n , but suppresses the image signal at the frequency (f _n -2f _pch ) (for a local oscillator with external synchronization for the lower side band). A low-noise amplifier 320 amplifies the signal. The amplified signal is mixed in the mixer 322 with the signal f _{g of the} local local oscillator supplied via bus 324 by the reconfigurable local local oscillator 326. This leads to the creation of signals (f _n + f _g ) of the sum and (f _n -f _g ) of the difference on the bus 328. Filter 330 intermediate frequency transmits a signal f _{pch of} intermediate frequency and attenuates frequencies outside the bandwidth of the modulated signal of interest. An analog-to-digital converter 332 operates by using the sync signal f _sync to obtain digital samples on a bus 344 with a frequency f _select . The down-converting digital converter 336 shifts the frequency, filters, and decimates the signal to obtain in-phase and quadrature signals of a lower sampling frequency on buses 338 and 340. Then, the digital signal processor 342 provides additional signal processing, including the coherent tracking described above, to obtain an output signal on bus 344 for output device 346.

An alternative method for evaluating both the phase reference signal and the IOSK, based on the reference subcarriers, is illustrated in FIG. As shown in FIG. 44, the complex “training” symbols carried by the reference subcarriers are entered on the bus 348, and the complex conjugate value of these symbols is taken, as shown in block 350. This complex conjugate value is multiplied by the known “training” sequence, fed via bus 352, by a multiplication block 354. This removes the modulation of the binary (± 1) clock sequence from the received “training” subcarriers by multiplying them by a synchronized, decoded, and differential transcoded clocked DPSK sequence. The resulting symbols supplied via bus 356 are processed by a finite impulse response filter 358 (FIR filter) to smooth out the resulting symbols in time, which provides a complex conjugate estimate of the local phase and amplitude on bus 360. This value is delayed by time delay block 362 and multiplied by an estimate of the reciprocal of the noise variance and supplied over the bus 364 by the multiplication unit 366. The noise variance is estimated by subtracting the smoothed estimate of the local phase and amplitude supplied over the bus 360 based on the input symbols (after suitable time alignment provided by the delay unit 368) at the summation point 370. Then the result is squared, as shown in block 374. The reciprocal value is approximated (with protection against division by zero), as shown in block 376. This weight of IOSK is interpolated by 18 subcarriers between pairs of adjacent “training” subcarriers, as shown by block 378, to obtain the resulting local IOSK weights on the bus 380. These IOSK weights are then used to multiply by the corresponding symbols carrying data and received on the bus 382, after which they are delayed appropriately, as shown in block 384. Then, the multiplication block 386 f forms a soft solution displayed on the 388 bus.

In Fig. 44, the buses along which the "training" symbols are transmitted are labeled EDUCA, and the buses through which data is transmitted are designated D. In addition, the filter 374 provides such a delay:

delay ≥ 1 / β, where β = 1/16,

and then

y _{n, m} = 2 · (1-β) · y _{n-1, m} - (1-β) ² · y _{n-2, m} + β ² · x _{n, m} .

These expressions refer to a bipolar IIR filter having a constant β time. The IIR filter calculates smoothed output samples “y” based on the input sample “x” and previous output samples.

Obtaining an IOSK weight combines amplitude weighting to sum the maximum ratios (QS) with phase correction to eliminate channel phase errors. This IOSK weight is dynamic in time and frequency and is estimated for each MPSK symbol.

- оценка комплексной сопряженной величины коэффициента усиления канала, и σ² - оценка дисперсии шума.Where

is an estimate of the complex conjugate value of the channel gain, and σ ² is an estimate of the noise variance.

Work according to the IOSK recovery method according to FIG. 44 involves capturing the frequency of the subcarriers and tracking it, as well as clocking the symbols as applied to the OFDM symbols. Methods of capturing frequency and clocking of symbols involve the use of cyclic prefix properties. Tracking the frequency and symbols is achieved by observing the phase drift from symbol to symbol depending on time or frequency (by subcarriers).

After capturing both the frequency and timing of the symbols, an attempt is made to synchronize with a combination of synchronization blocks of a clocking DPSK sequence by cross-correlating a differential-detected DPSK sequence with a combination of synchronization blocks. Differential detection is performed on all subcarriers under the assumption that the location of the “training” subcarriers is unknown at first. Cross-correlation of a known combination of block synchronization with detected bits of each subcarrier is performed. Subcarrier correlation is announced when a match of all 11 bits of a block synchronization combination is detected. Block synchronization (and the resolution of the subcarrier ambiguity problem) is established when the number of subcarrier correlations matches or exceeds threshold criteria (for example, 4 subcarrier correlations spaced into several subcarriers, the number of which is a multiple of 19 subcarriers).

After the synchronization of the blocks is established, the changing fields of the clocked DPSK sequence can be decoded. Differentially detectable binary bits of these variable fields are detected based on a “majority vote” of the “training” subcarriers, so decoding is possible when some of these subcarriers or binary bits are distorted. 16 blocks within each frame of the modem are given numbers from 0 to 15. Then the contents of the most significant binary digit (BSS) of the "block count" field are always set to zero, since the block count never exceeds 15. The modem frame synchronization is set in accordance with knowledge fields "counting blocks."

Coherent detection of this signal requires some coherent phase reference signal. The decoded information from the clocked DPSK sequence is used to remove modulation from the “training” subcarriers, leaving information about the reference signal of the local phase and noise. Turning to FIG. 44, we note that first, from the received “training” subcarriers, the modulation of the binary (± 1) clock sequence is removed, multiplying the mentioned subcarriers by a synchronized, decoded and differential transcoded clock DPSK sequence. To smooth the resulting symbols in time, use an FIR filter, which gives a complex conjugate estimate of the local phase and amplitude. This value is delayed and multiplied by an estimate of the reciprocal of the noise variance. The noise variance is estimated by subtracting the smoothed estimate of the local phase and amplitude from the input symbols (after proper time alignment), squaring and filtering the complex noise samples, followed by approximation of the inverse value (with protection against division by zero). This weight IOC is interpolated over 18 subcarriers between pairs of neighboring “training” subcarriers. The resulting local IOSK weights are then used to multiply by the corresponding local characters carrying the data.

In one embodiment, the low-pass filter 358 shown in FIG. 44 is an 11-tap FIR filter. This 11-tap FIR filter is used to dynamically estimate the complex coherent reference gain α at each subcarrier location for each symbol time. Temporal filtering using an 11-tap FIR filter and subsequent filtering by subcarriers is performed to calculate a local estimate of the coherent reference gain α for each location of the PPSK symbol in both time and frequency. A larger FIR filter with more taps could reduce the estimation error when the statistics of the signals are stationary, but the bandwidth would be too small to track Doppler-like changes in the signal at the maximum speeds typical of freeways. Therefore, 11-tap filters with a wedge-shaped symmetrical Gaussian impulse response are preferred. Instead of the IIR filter, a symmetric FIR filter is used because of its linear phase property, which gives a zero bias error for the (approximately) piecewise linear fading characteristic in the channel in the filter frequency range. This smoothed coherent reference signal output from the FIR filter is subtracted from the delayed input samples, which gives instant noise samples. These noise samples are squared and processed by the IIR filter 374, which gives an estimate of the noise variance, σ ² . This filter has a bandwidth that is narrower than the FIR filter, which allows a generally more accurate estimate of the noise variance. After appropriate sample delays, to match the filter delays, the weight α ^∗ / σ ² for each subcarrier is calculated. These values are smoothed and interpolated by subcarriers for each OFDM symbol to obtain more accurate estimates. This weight is special for each OFDM symbol, and each subcarrier provides a local (time and frequency) estimate and weight for the symbols forming the transition metrics for the subsequent Viterbi decoder.

The system illustrated in FIG. 44 was designed to match vehicles with fixed antennas. Numerous functions of reference subcarriers are described in relation to capture, tracking, estimation of channel state information (IOSC) and coherent operation. This system was designed for coherent operation in the FM broadcast band (88-108 MHz) with a fade bandwidth corresponding to vehicles moving at speeds characteristic of freeways. Various coherent tracking parameters are estimated using filters whose bandwidths approach the expected width (approximately 13 Hz) of the Doppler frequency band. With a stationary antenna, the corresponding tracking statistics for the input signal according to tracking algorithms are assumed to vary with a frequency not exceeding the Doppler frequency bandwidth.

Receivers HD Radio ^TM CDF standard can be used in conjunction with the antenna system with switch diversity. At the same time, the use of switching diversity antennas introduces sharp transients into the coherent tracking of a digital signal, which degrades the digital performance.

In the sense in which it is used in this description, the term "complex coherent reference gain (α)" for an MPSK symbol (depending on the location in time and frequency, since it is dynamic) is defined as α. This is complex a term that includes the real and imaginary components, which displays the gain and phase of the symbol associated with it. This value is estimated by the above processing and filtering. The term "composite reference signal x _n coherent channel" refers to value α of the composite signal for all reference subcarriers and for any time of one OFDM symbol.

In this example, it is assumed that the switching time characteristic of the diversity switching module is of the order of 10 microseconds, and the minimum response delay time is of the order of 10 milliseconds. The switching time of the order of 10 microseconds does not have a noticeable effect on the demodulated FM signal, and the overall improvement in diversity with frequency modulation can be significant, since signal fading is prevented. However, the switching transient negatively affects a coherently monitored digital signal. As shown in FIG. 23, the switching of the antenna module causes an intermittent transient in a coherent signal, with the signal itself immediately before or after the transient being approximately coherent. Filtering with a finite impulse response (FIR filtering) of the hopping transient distorts the estimated reference signal α of the coherent channel (as shown in FIG. 44), affecting the sampling range equal to the length of the FIR filter (11 taps). This distortion in the reference signal of the coherent channel increases the modulus of the noise samples by subtracting this reference signal from the input samples. Although noise increases only in the vicinity of the transient, the IIR filter expands this noise peak to the range of the IIR filter. This gives the effect of underestimating the noise variance when estimating in the immediate vicinity of the transient, while overestimating when estimating in the time range of the IIR filter in the nearest neighborhood of the transient. Both the distorted reference signal of the coherent channel and the noise dispersion errors contribute to the degradation of the digital signal. The invention provides a coherent tracking method that compensates for switching transients in an antenna diversity switching system.

Fig. 45 is a graph illustrating the effects of switching transients on noise estimates (SNR 20 dB). Line 438 displays the noise sample input to the IIR filter, and line 440 displays the filtered output. It can be seen that the input noise samples (squared) increase in the immediate vicinity of the transient. In fact, this is due to the distortion of the reference signal of the coherent channel due to the 11-tap FIR filter. Although the output of the IIR filter responds to noise peaks, these peaks undergo local suppression at the transient, but propagate throughout the IIR filter. These errors in the local estimation of noise variance contribute to the degradation of digital performance.

Fig. 46 is another graph illustrating the effects of switching transients on noise estimates (SNR 20 dB) in the case of a coherent reference signal fixed in accordance with the invention. Line 442 displays the noise sample introduced into the IIR filter, and line 444 displays the filtered output. Fig. 46 illustrates the same scenario as Fig. 45. However, in accordance with this invention, the reference signal of the coherent channel is adjusted. In FIR filtering or noise variance estimation nothing is changed.

Coherent tracking can be achieved on either side (in time) of the transient switching process, as shown in FIG. 24. On Fig presents a functional block diagram illustrating the generation of coherent reference channel signals. Demodulated complex symbols from several reference subcarriers are inputted on buses 450, 452 and 454. These complex symbols are combined with a conjugate 456 of a known reference sequence in mixers 458, 460 and 462 to remove data from the reference subcarrier symbols. FIR filters 464, 466, and 468 filter each complex subcarrier gain value to reduce noise. These FIR filters can be, for example, 7-tap filters. Block 470 indicates that the subcarrier gain values are smoothed over the reference subcarriers to further reduce noise in the estimation. This results in a coherent channel reference gain for each subcarrier on buses 472, 474, and 476. These coherent channel reference gains are added at the summation point 478 to obtain a composite coherent channel reference signal x _n on bus 480.

Coherent estimation filters cannot use samples of signals that span the transient on both sides, so tracking filters can only use signal samples up to the transient (not inclusive) from any direction in time. The instant of occurrence of the transient signal must be identified. Although (theoretically) there should be the possibility of implicit communication of the switching moment to the receiver’s modem, this may turn out to be practically impossible. For the modem, reliable transient detection within the framework of coherent tracking algorithms is preferred. Coherent tracking algorithms can be modified in the vicinity of the transient to provide approximately coherent tracking in the immediate vicinity of the transient, with the exception of the symbol where the transient occurred. Needless to say, an error in the estimation of noise variance arises from an error in the reference signal of the coherent channel. Therefore, if the reference signal of the coherent channel is accurate, then there is no need to change the estimate of the noise variance.

It is required to ensure the detection of "blindly" spasmodic transient due to the switching of antenna elements. A feature related to the detection of “blind” is related to the method, in which the algorithm provides viewing of demodulated modem symbols and does not have direct knowledge of the moment of switching in the diversity switching module. The method involves processing tracking samples of a smoothed coherent reference signal (complex filtered values of α), which are indicated by the letter "x" in Fig. 47 and in this description, at a repetition rate of OFDM symbols (i.e., approximately 344.5 OFDM symbols per second in one example). The values of the complex channel gain α are grouped by all reference subcarriers to obtain one composite complex value “x” for each OFDM symbol.

It is assumed that the transient (which can last approximately 30 microseconds) is much shorter than the time taken by the symbol, and that this transient occurs within only one symbol. Symbols on either side of the transient are not significantly distorted by this transient and can be used for coherent tracking and evaluation. However, characters whose distance from the transient is within half the range of the FIR filter (for example, 5 samples for an 11-tap FIR filter or 3 samples for a 7-tap FIR filter) are filtered to evaluate a coherent reference signal. This is explicitly illustrated in FIG. 24, which shows the real and imaginary components of the complex “x” values (solid lines in the graph), and the number of taps of the FIR filter was reduced from 11 to 7 to minimize the range of transient effects. Reducing the number of taps of the FIR filter to 7 has the effect of a slight decrease in the gain of the filter during noise, but improves the transient response.

On Fig shows the distortion of coherent tracking within ± 3 characters from the transition process, in which the coherent reference signal makes the transition from the pre-transition value to the post-transition value for α. One way to detect a transient involves testing the hypotheses of the “x” samples on either side of the symbol in question, where the transient should be detected. In other words, if it is necessary to detect a transient at the location n of the symbol, then we observe samples x _n-1 and x _{n + 1} . Hypothesis testing continues for each subsequent location of the symbol, and it is also assumed that observations of past and future samples with respect to sample “x” can be made throughout the entire suitable delay in signal processing algorithms, which can be adjusted later. It is also assumed that a transient is detected when the values of the samples x _n-1 and x _{n + 1} turn out to be quite different (or incoherent). Several forms of detecting this difference are analyzed, modeled and compared in the channel with fading. The simple difference (x _n-1 -x _{n + 1} ) of complex values or the modulus | x _n-1 -x _{n + 1} | This difference is insufficient for the localization of the transient process, since this difference is scaled by the signal module at this moment, and the peak of the difference is not displayed. However, this difference can be scaled inversely with its modulus after detecting a peak to check if a transient is present. Although various forms of detection criteria may be effective, this example uses squares of modules instead of modules to do without calculating square roots. The following three-stage algorithm for detecting the location of the transient process has proven to be effective in a noisy channel with fading and relatively simple when calculating the sequence of samples x _n .

Since the expressions for peak detection and transient use the future values of the “x” samples, it is convenient to calculate the difference “different squared” before calculating the next two expressions. A typical threshold value is 0.05. An example of a transient detection output signal is shown in FIGS. 48 and 49. FIG. 48 shows the results of applying a transient detection algorithm for 800 OFDM symbols. On Fig shows the results of applying the transient detection algorithm in the vicinity of the symbol under the number 200. These drawings show the values of the real part 484 and the imaginary part 482 of the filtered reference signal α of the coherent channel and the results of the detection of the transient process, showing pulses 486 detected transient processes. Fig. 49 shows, on an enlarged scale, the range near the symbol number 200. Note that a transient was not detected in the symbol number 600 because the threshold is insensitive to small differences during the transient that have minimal impact on performance. However, the value of this threshold works well up to an SNR of 0 dB.

Now that the transient can be reliably detected, the next step is to adjust the samples of the α coherent reference signal in the immediate vicinity of the transient. A relatively simple solution is to discard the values of the samples α near the transient, which include the symbol where this transient is detected. These values are replaced by the closest sample value α, which was not exposed to the IIR filter (it is located at a distance of 4 samples from the transient in the case of a 7-tap IIR filter). This gives the result displayed by the dashed lines of the graphs in FIG. An example of a sequence in the vicinity of the detected transient is presented in Table 3. Table 3 shows the input values of the samples α and the aligned output values of the samples α when the transient is detected at the symbol location n. Input samples are ordered as characters. But the aligned output sequence for α samples is adjusted to minimize distortion introduced by the IIR filter, as described previously.

Table 3
Correction of the complex channel α gain (for each reference subcarrier) in the vicinity of the transient detected at symbol location n

To achieve the results presented in table 3, you can use a simple algorithm with a suitable point and buffering for the range of "x" samples. In the case of using a 7-branch IIR filter to obtain α samples, the operation is as follows. If the transient is detected within 3 characters before the current character (for example, between place n-3 and place n-1), then instead of using the current value of the sample α, the value (α _n-4 ) of the sample α, which is located in 4 characters before the transient. Similarly, if a transient is detected within 3 characters after the current character (for example, between location n + 1 and location n + 3), then instead of using the current value of the sample α, the value (α _{n + 4} ) of the sample α is used, which is in 4 characters after the transient. If the current symbol is in the location (for example, n) of the detected transient, then the average value is used - (α _n-4 + α _{n + 4} ) / 2 - samples α, which are within ± 4 characters on either side of the transition process. Otherwise, the current input values of the samples α are used (for example, in places from n-5 to n-4 and from n + 4 to n + 5). Of course, the details of this algorithm can be adjusted to fit a different IIR filter range. A functional diagram illustrating the correction of α values in the vicinity of the detected transient is shown in FIG. 50. FIG. 50 shows that an uncorrected coherent channel gain for each subcarrier is input across buses 500, 502, and 504. These input signals are summed at summation point 506 to obtain a composite reference of a coherent channel on bus 508. Transient detector 510 detects transients processes in the composite reference signal of a coherent channel. The uncorrected coherent channel gains are then adjusted in the vicinity of the transient, as shown in block 512. This gives a corrected coherent channel gain for each reference subcarrier on buses 514, 516, and 518.

Analysis and modeling of algorithm improvements show that they work well enough for analyzed and simulated cases. These cases include a channel with smooth and selective fading and with a Doppler frequency bandwidth corresponding to the speeds typical of freeways and such insignificant noise as the corresponding 0 dB SNR. However, other conditions in the channel should be considered. For example, impulse-like noise can lead to false transient detection. In this case, the adjusted values of the “x” samples of the coherent reference signal are suitable. However, the estimate of noise variance will be distorted. The noise impulse can be very high for the symbol (s) where this impulse originated, but the IIR filter can suppress this noise estimate value at the moment the impulse occurs and extend the noise estimate from the reaction time to the impulse disturbance of the IIR filter. In this case, it would be preferable to directly transmit the loud noise samples parallel to the IIR filtering path (with suitable delay matching). For characters in which the noise pulse is significantly larger than the output of the IIR filter, this noise pulse should be used to determine the estimated noise variance for these symbols. When using the direct transmission path for these noise pulses, the energy entering the IIR filter for these symbols should be reduced so that the local noise peak does not propagate over the IIR filter range. It is easy to envision several variations of this process for processing noise peaks in the estimation of noise variance. One such modification of the estimation of noise variance, providing matching noise-like noise, is presented in Fig. 51. On Fig shows that the input signals for each reference sample are supplied on the bus 520 and added to the reference sample α ₀ in the adder 522. The noise variance estimates are filtered by the reference subcarriers to reduce the estimation errors, as shown in block 524. The filtered variance samples then they are used to calculate the noise variance, as shown in block 526. This results in a variance of “disp0 _n ” on bus 528 (where dis0 _n = min (2 ∗ disPHI _n-1 + ε, disp _n )) and the variance “disp1 _n "On the bus 530 (where dis1 _n = max (0.5 * dis _n -dISBR _n-1 )). The dispersion “disp0 _n ” is then filtered at block 532, and “dispers1 _n ” is delayed, as shown by block 534. The filtered “disp0 _n ” and delayed “disp1 _n ” are summed in the adder 536 to obtain an estimate of the noise variance output via bus 538. The output dispersion "dispersion _n-1 " of the filter 532 is fed back to block 524 via bus 540.

The transient detection method described for switching an antenna also works in the case of impulse noise. In contrast to the case of switching with a transient, in which the value of the noise of the transient is small, the impulse noise can be large. In addition, sometimes the switching transient is undetected, which causes an erroneous (albeit small) value of the α sample, but causes increased noise in the vicinity (± 3 characters), which has an effect similar to that exerted by impulse noise. In these cases of impulse-like noise, noise dispersion estimation filters smooth out the value of impulse noise over many (for example, 64th) symbols, so that the noise at the moment of impulse noise is underestimated when estimated, while noise within ± 32 OFDM symbols is overestimated when estimated due to the “blurring” of the noise sample. This causes a deterioration in the quality of soft symbols supplied to the Viterbi decoder. Therefore, this regulation can be introduced into the estimation of noise variance in order to improve the performance for these cases of impulse noise. A new estimate of the noise variance consists of the sum of the long-term variance of the Gaussian noise and the short-term variance of the pulse-shaped noise.

Turning to FIG. 51, we note that the value “disp0” is a normal sample of the noise variance introduced into the IIR filter for the subsequent evaluation of the noise variance estimate. Samples “disp” are sorted to represent normal squared samples (“disp0”) Gaussian noise and squared samples (“disp1”) pulse-shaped noise. An IIR filter estimates the variance of normal Gaussian noise, while a delay path without filtering provides a direct supply of pulsed noise variance samples. When samples of “disp” noise are within the appropriate range of the current IIR filter output signal, the value “disp0” is set to “disp”, and “disp1” = 0. When the noise value is unusually high (compared to the current output signal of the filter), then it is more likely to sample pulse-shaped noise, which should not fall into the long-term filter. In this case, the value of “disp0” is limited, although some of the redundant values are placed in the element “disp1”. The value of "disp1", respectively, is delayed to match the delay of the IIR filter. The result is that the samples of the impulse noise are properly estimated at the right moment, and this noise is not spread across many samples, as shown in FIG. 52, which gives an improvement over the results shown in FIG. 46, where the impulse noise is obtained around the symbol number 600.

52 is a graph illustrating the effects of switching transients on noise estimates (SNR 20 dB) when a noise variance modification is applied. Line 550 displays the noise variance of the symbols, while line 552 displays the filtered output. Note that the missed transient of the symbol at number 600 is properly adjusted in accordance with the increased noise dispersion in the immediate vicinity of the symbol at number 600.

The invention also provides a method for estimating character noise variance when this noise may include pulsed noise samples among nearly Gaussian noise samples. Nonlinear filtering of the squares of the error samples is provided, the nonlinear filtering method includes the steps of calculating the squares of the error samples between the values of the coherent reference signal and the values of the new symbols, and using the nonlinear filtering method, in which normal (squared) Gaussian noise samples are passed through a filter (for example, an FIR or IIR filter) to estimate the dispersion of Gaussian noise, while adding (squared) samples of the pulse-like noise and a filter output signal to produce a new noise variance displaying sum long averaged Gaussian-like noise variance, and short-term impulsive noise variance.

The method may further include sorting the squared Gaussian noise samples and the pulsed noise samples, determining whether each new input (squared) noise sample is pulsed, by comparing with a multiple of the current average noise dispersion estimate output from the filter ( plus a constant for taking into account jump-like transients), and if a squared noise sample is not defined as pulse-shaped, then this value is introduced into and polzuemy filter to estimate the long-term dispersion gausobraznogo noise.

This invention provides improvements to the coherent tracking algorithms that are used autonomously with a blind-switched antenna diversity switching system. These same enhancements can also moderate degradation due to impulse noise or non-Gaussian noise, for example, from an adjacent frequency-modulated analogue noise source.

Although the invention has been described with a few examples, it will be apparent to those skilled in the art that, within the scope of the claims defined in the following claims, it is possible to make changes to the described embodiments.

Claims (32)

1. A method for coherently tracking a radio signal including a plurality of reference subcarriers subjected to digital modulation, including the steps of receiving symbols transmitted on the reference subcarriers using a plurality of antenna elements, wherein switching between the antenna elements causes a jumplike transient , combine the symbols of the reference subcarriers with the conjugate value of the known reference sequence to obtain multiple samples, median filtering of the samples d I obtain filtered samples and smoothed samples for each reference subcarrier in the set of reference subcarriers to produce a coherent reference signal estimate for each of the subcarriers. 2. The method according to claim 1, further comprising the steps of using the aforementioned set of samples and a coherent estimate of the reference signal for each of the subcarriers to obtain a noise dispersion signal, filtering the noise dispersion signal to obtain a filtered noise dispersion signal, choosing between the dispersion signal noise and a filtered noise dispersion signal to determine the selected noise dispersion signal and smooth the selected noise dispersion signal for each of the reference subcarriers on the set of reference subcarriers for I am deriving an estimate of the signal noise variance for each of the subcarriers. 3. The method of claim 2, wherein the step of using the plurality of samples and coherently estimating the reference signal for each of the subcarriers to obtain a noise dispersion signal includes the steps of subtracting the coherent estimate of the reference signal from said plurality of samples to obtain a noise sample and constructing squared noise sampling module to produce a noise dispersion signal. 4. The method according to claim 2, wherein said step of filtering the noise dispersion signal to obtain a filtered noise dispersion signal includes the step of sequentially passing the noise dispersion signal through a median filter and an additional filter for smoothing the noise dispersion signal in frequency. 5. The method according to claim 4, wherein the additional filter comprises a filter with an infinite impulse response or a filter with a finite impulse response. 6. The method according to claim 2, further comprising scaling the noise dispersion signal before the selection step between the noise dispersion signal and the filtered noise dispersion signal to determine a selected noise dispersion signal. 7. A receiver for coherently tracking a radio signal including a plurality of digitally modulated reference subcarriers, comprising a plurality of antenna elements for receiving a radio signal, wherein switching between the antenna elements causes a jumplike transient, an input for receiving a radio signal, and a processor for coherently tracking the radio signal by receiving the symbols transmitted on the reference subcarriers, combining the symbols of the reference subcarriers with the conjugate value of the known reference sequence and to obtain a plurality of samples, median filtering of the samples to produce filtered samples, and smoothing the samples for each reference subcarrier in the set of reference subcarriers to produce a coherent reference signal estimate for each of the subcarriers. 8. The receiver according to claim 7, in which the processor further uses said plurality of samples and a coherent estimate of the reference signal for each of the subcarriers to obtain a noise dispersion signal, filters the noise dispersion signal to obtain a filtered noise dispersion signal, selects between the noise dispersion signal and the filtered signal noise dispersion for determining a selected noise dispersion signal and smooths the selected noise dispersion signal for each of the reference subcarriers on the set of reference subcarriers to obtain signal noise dispersion quotes for each of the subcarriers. 9. The receiver of claim 8, wherein the processor further subtracts the coherent estimate of the reference signal from said plurality of samples to obtain a noise sample, and squares the noise sample module to obtain a noise dispersion signal. 10. The receiver of claim 8, wherein the processor sequentially passes the noise dispersion signal through a median filter and an additional filter. 11. The receiver of claim 10, wherein the additional filter comprises a filter with an infinite impulse response or a filter with a finite impulse response. 12. The receiver of claim 8, wherein the processor scales the noise dispersion signal before the selection step between the noise dispersion signal and the filtered noise dispersion signal to determine a selected noise dispersion signal. 13. A method for coherently tracking a radio signal including at least one reference carrier, digitally modulated, including the steps of using a plurality of antenna elements to receive a radio signal, wherein switching between the antenna elements causes a jumplike transient, demodulate the reference carrier to obtain complex coherent reference gain, detect a transient that affects the complex coherent reference gain filter the complex coherent reference gain factors to obtain smoothed complex coherent reference gain factors and process the smoothed complex coherent reference gain factors, discarding the values of the complex coherent reference gain factors closest to the transient, and replacing the values of the complex coherent reference gain factors closest to the transition process , the closest value of the complex coherent reference gain factors at which e does not affect the filter. 14. The method according to item 13, in which the radio signal contains many reference subcarriers, and which regulate the complex coherent reference gain in the vicinity of the transient for each of the reference subcarriers. 15. The method according to item 13, in which the step of detecting a transient process that affects the complex coherent reference gain factors includes the steps of processing the complex coherent reference gain factors for multiple reference subcarriers and group the complex coherent reference gain factors for all reference subcarriers to obtain one composite reference coherent channel signal for each OFDM (orthogonal frequency division multiplexed) symbol (OFDM symbol). 16. The method according to item 13, in which the step of detecting a transient process that affects the complex coherent reference gain factors includes the steps of calculating the modulus of the difference of the plurality of composite reference signals of the coherent channel, determining whether the module is a local peak, and if the module is a local peak, then the module is scaled back by the sum of the module of the difference. 17. The method according to item 13, in which the step of detecting a transient that affects the complex coherent reference gain, includes the steps of calculating the sequence of samples x _n based on the complex coherent reference gain, calculate the square of the difference diffsg _n (different squared _n ) between samples x _{n + 1} and x _n-1 , set the peak detection variable detpeak _n (detected_peak _n ) to unity if diffsg _n ≥diffsg _n-1 (different squared _n ≥dif square _n-1 ) or diffsg _n ≥diffsg _{n + 1} (different square _n ≥ different square _{n + 1} ), otherwise set the peak detection variable to 0, and indicate the presence of a transient if

(different squared _n / (| x _{n + 1} | ² + | x _n-1 | ² )) is greater than the predetermined threshold value. 18. The method of claim 13, wherein the processing step of the complex coherent reference gain factors includes discarding values (α) of the complex coherent reference gain factors close to the transient, which include the symbol where the transient is detected , and replace the discarded values (α) with the nearest value α, which is not affected by the filter used to estimate α. 19. The method according to item 13, in which the processing step of the complex coherent reference gain includes steps in which if the distorted value (α) of the complex coherent reference gain is detected within 3 characters before the current symbol, then instead of using the current α values use the value (α _n-4 ), which is 4 characters before the transient, if the distorted value (α) of the complex coherent reference gain is detected within 3 characters after those character, then instead of using the current value of α, use the value (α _{n + 4} ), which is 4 characters after the transient process, if the distorted value (α) of the complex coherent reference gain is at the location of the current detected symbol, then use the average value (α _n-4 + α _{n + 4} ) / 2 of the values of α samples that are in the position of ± 4 characters on either side of the transient, and if the distorted value (α) of the complex coherent reference gain is not detected but, then use the current input values of α. 20. A receiver for coherent tracking of a radio signal, including at least one reference carrier, digitally modulated, containing a plurality of antenna elements connected to the input, while switching between the antenna elements causes an abrupt transition process, an input for receiving a radio signal and a processor for demodulating a reference carrier to obtain complex coherent reference gain factors, for detecting a jumplike transient that affects complex coherent s reference gains, and for adjusting the complex coherent reference gains in the vicinity of the transient to produce adjusted coherent reference gains. 21. The receiver according to claim 20, in which the processor substitutes the previous or future value of the complex coherent reference gain in place of the complex coherent reference gain in the vicinity of the transient. 22. The receiver of claim 20, wherein the processor filters the complex coherent reference gain factors and processes the smooth complex coherent reference gain factors, discarding the complex coherent reference gain factors closest to the transient, and replacing the values complex coherent reference gain factors closest to the transient, the closest value of complex coherent supports gain factors not affected by the filter. 23. The receiver according to claim 20, in which the radio signal contains many reference subcarriers, and which regulate the complex coherent reference gain in the vicinity of the transient for each of the reference subcarriers. 24. The receiver of claim 20, wherein the processor processes the complex coherent reference gains for the plurality of reference subcarriers and groups the complex coherent reference gains across all the reference subcarriers to obtain one composite coherent channel reference for each OFDM symbol. 25. The receiver of claim 20, wherein the processor calculates a module of the difference of the plurality of composite reference signals of the coherent channel, determines whether the module is a local peak, and if the module is a local peak, then the module is scaled back by the sum of the module of the said difference. 26. The receiver of claim 20, wherein the processor receives a sequence of samples x _n, calculates a difference square diffsg _n (razn.kvadr. _N) between the samples x _{n + 1} and x _n-1, sets the variable detection ditpeak _n peak (detected. _peak _n ) equal to one if diffsg _n ≥diffsg _n-1 (different square _n ≥ different square _n-1 ) or diffsg _n ≥diffsg _{n + 1} (different square _{n n} ≥ different square _{n + 1} ), otherwise set the peak detection variable to 0, and indicate the presence of a transient, if

(different squared _n / (| x _{n + 1} | ² + | x _n-1 | ² )) is greater than the predetermined threshold value. 27. The receiver according to claim 20, in which the processor discards the values (α) of the complex coherent reference gain factors close to the transient, which include the symbol where the transient is detected, and replaces the discarded values (α) with the nearest value of α, by which is not affected by the filter used to estimate α. 28. The receiver according to claim 20, in which if the distorted value (α) of the complex coherent reference gain is detected within 3 characters before the current symbol, then instead of using the current value of α, the value (α _n-4 ), which is located in 4 symbols before the transient, when a distortion value (α) complex coherent reference gain is detected within 3 symbols after the current symbol, the value (α _{n + 4)} is used instead of using the current value of α, which tries ditsya in 4 symbols after the transient, if a distortion value (α) complex coherent reference gain is at the location of the current detected symbol, then using the average value (α _{n-4 +} α _n + ₄₎ / 2 from the values α samples which are at the position of ± 4 characters on either side of the transient, and if the distorted value (α) of the complex coherent reference gain is not detected, then the current input values are used. 29. A receiver for coherent tracking of a radio signal, including at least one reference carrier, digitally modulated, containing a plurality of antenna elements connected to the input, while switching between the antenna elements causes an abrupt transition process, an input for receiving a radio signal and means for demodulating the reference carrier to obtain complex coherent reference gain factors, for detecting a transient that affects complex coherent reference coefficients gain factors, and for regulating complex coherent reference gain factors in the vicinity of the transient to obtain adjusted complex coherent reference gain factors. 30. A method for estimating the noise variance of symbols in a radio signal, when the noise may include pulsed noise samples among Gaussian noise samples, the method including the steps of summing the input samples and coherent reference samples to obtain error samples, calculating the squares of the error samples divide squared samples of Gaussian noise and squared samples of impulse noise, and perform non-linear filtering of squares of samples of errors to obtain an estimate of the variance w Ma, representing the sum of long-term-averaged Gaussian-like noise variance, and short-term impulsive noise variance. 31. The method of claim 30, comprising determining whether each squared noise sample is pulse-like by comparing the sample with a multiple of the current average noise variance estimate, and if the squared noise sample is not defined as pulse-shaped then a squared noise sample is introduced into the filter used to estimate the long-term dispersion of Gaussian noise, and if the noise sample is defined as pulse-shaped, then the squared noise sample in the filter is reduced for I prevent reassessment of long-term gaussian noise variance. 32. The method of claim 30, comprising creating an excess noise value including a portion of the reduced squared noise sample and adding the excess noise value to the Gaussian noise variance estimate to obtain a composite noise variance representing the sum of the long-term averaged dispersion of gaussian noise and short-term dispersion of impulse noise.

Images