HK1158403B - Ofdm frame synchronisation method and system - Google Patents

Description

OFDM frame synchronization method and system

Technical Field

The present invention relates to a method, system and computer program for OFDM synchronization, and more particularly to OFDM frame synchronization.

Background

Orthogonal Frequency Division Multiplexing (OFDM) subdivides a high data rate input data stream into a number of parallel data rate reduced sub-streams, with each sub-stream being modulated and transmitted simultaneously on separate orthogonal sub-carriers. Referring to fig. 1, an OFDM transmitter 10 includes a symbol mapper 16, the symbol mapper 16 combining input serial data 14 to form a symbol. The symbols are modulated with baseband subcarriers by an inverse Discrete Fourier Transform (DFT)18 and then serialized to form temporary OFDM symbols. The cyclic prefix is formed of several samples selected from the tail end of the temporary OFDM symbol. The cyclic prefix is connected to the beginning of the corresponding temporary OFDM symbol. The cyclic prefix and the temporary OFDM symbol together form an OFDM symbol, wherein the cyclic prefix forms the beginning of the OFDM symbol and the temporary OFDM symbol forms the remainder. The OFDM symbols are then transmitted to a digital-to-analog converter (DAC)20 where they are converted to analog form.

Before transmitting the first OFDM symbol, the transmitter 10 transmits a special signal called a preamble for synchronization purposes. Thus, an OFDM frame includes a preamble signal followed by a plurality of OFDM symbols. Upon receipt of the OFDM frame, the OFDM receiver 24 performs the inverse operation of the OFDM transmitter 10 in reverse order. However, the system clock of the receiver 24 must be synchronized with the transmitter 10 before any receiver algorithm can be employed. Symbol timing refers to the processing operations used to determine the precise time instants at which each OFDM symbol begins and ends. This time instant is used to locate the DFT window of the receiver (i.e., the set of samples used to compute the DFT of each received OFDM symbol) and thereby demodulate the subcarriers of that received OFDM symbol. While upper layer OFDM protocols (e.g., OFDM Medium Access Control (MAC) policies) provide some rough guidance to the start of an OFDM symbol, and do not provide an exact indication thereof. In addition, the MAC protocol in the receiver can only be operated if the received OFDM symbols have been previously synchronized and decoded; since the synchronization mechanism of the MAC layer is more focused on tracking fluctuations of the reference clock signal.

Conventional synchronization methods rely on the detection of a preamble. Referring to fig. 2, the preamble includes a short OFDM symbol (or preamble symbol) 30 used only in the preamble signal. Specifically, the preamble includes only a set of samples obtained from the output of a short Inverse Fast Fourier Transform (IFFT); but not the cyclic prefix. The preamble symbols are typically shorter than the OFDM symbols used in the remaining OFDM frames. The use of short preamble symbols minimizes the overhead (for overall transmission efficiency) of transmitting the preamble; and can simplify preamble implementation. The problem of symbol synchronization can be divided into two steps, namely:

-timing synchronization, comprising determining a time shift (shift) between a transmitted preamble symbol and a receiver DFT window; and

frame synchronization, which consists in determining the starting point of the payload (or of the last symbol in the preamble) of the received OFDM signal.

Timing synchronization can be obtained by signal correlation in the time domain (t.m.schmidl and d.c.cox, IEEE trans.oncommun., 1997(45), 1613-. Phase correlation involves determining the phase shift between the training DFT and the preamble symbols from the cross-correlation peak. The phase shift may be represented by an angular rotation, where the magnitude of the angle provides an indication of the range of the phase shift. Phase correlation provides better performance in the presence of strong narrowband-to-interband interference. In particular, since the preamble symbols remain the same during timing synchronization, even for signals with lower valued (negative) signal-to-noise ratio (SNR), averaging the number of symbols allows the DFT window to be aligned. Once timing synchronization is obtained, it is assumed that the DFT window of the receiver is aligned with each preamble symbol.

The frame sync successively correlates successive preamble symbols to detect a last preamble symbol, wherein at least one last preamble symbol is typically inverted (sign inverted). The formation of the correlation may be performed in the time domain (i.e., before the DFT) or in the frequency domain (i.e., after the DFT). The correlation process is based on the observation that the maximum correlation value is obtained if two consecutive preamble symbols are the same. However, if the symbols are of opposite sign, the minimum correlation value is obtained. Thus, in use, the associated output is examined for mutations therein. However, when the signal is highly corrupted by noise resulting in a negative signal-to-noise ratio (SNR), the preamble symbols cannot be processed in this way because the positions of the inverted symbols are lost (k.shi, e.serpin, ieee trans.on commun., 2004, 3(4), 1271-.

The repetition structure has been included in the preamble by the ieee802.11a/Hyper-LAN-II standard (ieee802.11a "part 11: wireless Local Area Network (LAN) Medium Access Control (MAC) and physical layer (PHY) specifications: high speed physical layer in the 5GHz band, month 7 1999, and ETSI DTS/BRAN 030003-1," broadband wireless access network (BRAN); hyterlan type 2 function specifications. part 1-Physical (PHY) layer, month 6 1999). More specifically, these standards form preambles such as [ S S S S S S S S L L ] with a series of short symbols (S) followed by two long symbols (L). Wireless Local Area Networks (WLANs) (ieee802.11a) define short symbols as the output of short FFTs (i.e., having a smaller number of points than the FFT used for the data symbol), but do not include a cyclic prefix. Similarly, a long symbol is defined as the output of the same FFT used on the data symbols, but does not include a cyclic prefix. Both long and short symbols are fixed according to the standard.

The long and short symbols are used for fine time/frequency synchronization and channel estimation. More specifically, short symbols are used for timing alignment and long symbols are used for frame synchronization. However, HiperLAN is designed to operate in a signal state of positive SNR; it is also difficult to use this method for synchronization with signals with low SNR values because the reliability of the synchronization is highly deteriorated (i.e. there is a high probability that the synchronization will not be correct). Similarly, the HomePlug-AV system (home plug-in powerline alliance, "HomePlug AV baseline specification" version 1.1, 5 months 2007) employs a preamble in the form of S-S S.

In this case, frame synchronization is obtained by looking for a negative sign in the preamble. However, these methods have been designed to operate in signal conditions of positive SNR; and is difficult to use for synchronization with signals of low SNR values.

Disclosure of Invention

The invention provides a method for synchronizing OFDM frames, which comprises the following steps:

(a) receiving a transmitted OFDM frame containing a preamble, the preamble comprising a predetermined number of preamble symbols, the preamble symbols to form a known preamble codeword;

(b) cross-correlating (cross-correlating) successive pairs of symbols in the received frame to generate a cross-correlation output;

(c) selecting a codeword from a plurality of predetermined codewords that most closely matches a selected number of cross-correlation values from the cross-correlation output;

(d) comparing the selected codeword to the preamble codeword;

(e) in the event that the selected codeword matches the preamble codeword, deciding that synchronization has been obtained; and

(f) in case the selected codeword does not match the preamble codeword, performing the following steps:

-obtaining additional symbols;

-cross-correlating the additional symbols with a previously received symbol (immediate preceding received symbol) to generate a further cross-correlation value;

-connecting the further cross-correlation value to the cross-correlation output;

-removing the inverse (opposing) cross-correlation value from the cross-correlation output; and

-repeating steps (c) to (f) until a predetermined stop criterion is reached.

The invention has the beneficial effects that:

error correction codes are currently used in channel coding to reduce the bit error rate in data communications. In contrast, the present invention uses error correction codes for synchronization. In particular, the present invention uses an error correction code to compensate for errors in a received synchronization pattern (pattern).

A particular advantageous aspect of the present invention is the use of short symbols in the preamble for synchronization. In particular, it is possible to use longer symbols in the preamble (making the lower SNR signal more robust), however, this approach has some disadvantages. In particular, the generation of a longer preamble signal increases the overhead of the synchronization process in the overall transmission efficiency of the communication system. In addition, timing synchronization becomes considerably more complex. Similarly, using averaging can result in simpler timing synchronization and robustness (robustness) of lower SNR signals, but averaging cannot be used for frame synchronization (because the timing reference needed to perform frame synchronization is lost when averaging improves SNR).

In contrast, the invention can make the frame synchronization under the lower SNR value have robustness under the condition of using shorter symbols. In addition, the present invention may improve the reliability of frame synchronization without using symbol averaging. In particular, the present invention utilizes error correction codes for frame synchronization.

Drawings

Embodiments of the invention are described below with reference to the accompanying drawings, in which:

FIG. 1 is a block diagram of an OFDM transmitter in communication with an OFDM receiver;

FIG. 2 is a block diagram of a preamble in an OFDM stream;

FIG. 3 is a block diagram of a system of the preferred embodiment; and

fig. 4 is a flow chart of a method of the preferred embodiment.

Detailed Description

Common to some prior art methods is that the preferred embodiment performs frame synchronization by correlating consecutive symbols in the received preamble. However, the preferred embodiment then converts the resulting cross-correlation values into binary sequences. More specifically, the preferred embodiment converts the maximum correlation value to a value of "1" and converts the minimum correlation value to a value of "-1". As discussed above, when the OFDM signal is corrupted by noise, the cross-correlation output may be any value between the maximum and minimum values. In this case, a threshold value for determining whether to convert a given cross-correlation value into a binary "1" or a binary "0" is set at a midpoint between the maximum value and the minimum value. Alternatively, two thresholds may also be defined such that some numbers with values close to 0 are not assigned bits (in which case the bit values are unknown to the decoder). Furthermore, the system may use a soft decoding process that uses the cross-correlation output value itself as an input to the decoder.

In parallel with (and in support of) the above processing, the preferred embodiment embeds a predetermined error correction code in the preamble. In particular, the preferred embodiment replaces the preamble of a conventional OFDM frame with an encoded preamble consisting of a set of preamble symbols whose signs (signs) are defined such that the binary sequences resulting from the above described process (converting the output of the cross-correlation process into binary form) are the codewords of the error correction code. In the case where the received signal is very noisy, the output of the correlation process is highly degraded; and the binary sequence obtained therefrom is corrupted. However, the known error correction code contained in the orthogonally transmitted preamble enables the detection and correction of corrupted bits. In addition, once the complete error correction code is received, frame synchronization is obtained.

More specifically, the preferred embodiment employs a preamble divided into two portions. The first portion includes a plurality of short symbols having the same sign (i.e., [ S.. S ]). The second part of the preamble comprises a predetermined number (N) of short symbols, the sign of which is determined by the bit sequence [ a (0) a (1.. a (N-1) ], wherein a (i) may be +1 or-1. In other words, the second part of the codeword comprises the symbols [ a (0) · S a (1) · Sa (2) · S.. a (N-1) · S ].

The bit sequence a (i) is established using the following mechanism:

selecting a codeword of length N from a set of codewords of a given error correcting code, wherein the bits of the selected codeword are denoted [ b (0) b (1.. b (N) ], a given bit b (i) can take either a value of +1 or-1;

setting a first symbol a (0) in the second portion of the preamble to match a first bit in the selected codeword (i.e., setting a (0) ═ b (0)); and

the remainder of the notation a (i) is set according to the recursive expression a (i) ═ b (i) × a (i-1), i ═ 1, 2.

Thus, the signs of the short symbols in the second part of the preamble are effectively set such that repeated cross-correlation of pairs of consecutive short symbols generates the selected codeword. It will be appreciated that cyclic codes are particularly useful in the preferred embodiment, as one of the main properties of cyclic codes is that the cyclic shifts of codewords are also codewords. It should be understood, however, that the preferred embodiments are not limited to cyclic codes, and that other types of codes may be used instead.

Referring to fig. 4, the receiver 70 of the preferred embodiment includes a DFT module 72, which DFT module 72 continuously applies a training DFT to the input signal. The symbols from successive DFT operations are sent to a cross-correlation module 74, and the cross-correlation module 74 calculates the cross-correlation between the symbols. In particular, correlation module 74 outputs a positive maximum value (e.g., +1) indicating two consecutive signs of perfect match, outputs 0 indicating an uncorrelated sign (i.e., noise), and outputs a negative minimum value (e.g., -1) indicating two matching consecutive signs of an odd sign.

The output of the cross-correlation module 74 is sent to a decoder 76, which decoder 76 may employ any type of decoding process, including hard decoding and soft decoding. The decoder 76 computes hamming distance (or other distance metric) between the output from the cross-correlation module 74 and a set of known codewords (a set of known, fixed hamming distances). Specifically, the decoder 76 matches the output from the cross-correlation module 74 with the most similar of its codewords; and outputting the correlated matched codeword. This method is employed in the case where the transmitted preamble has a codeword that may have been distorted by noise on the channel or the like. Thus, the decoder 76 effectively attempts to correct this distortion. The output codeword is sent to a decision block 78 where the output codeword is compared to the known preamble codeword. And if the output code word is matched with the lead code word, judging that the synchronization occurs. Otherwise, a training DFT is applied to the next input signal and the above processing steps are repeated until synchronization is achieved.

Referring more specifically to fig. 4, when a hard decoder is used, the preferred embodiment includes the steps of:

(a) receiving an OFDM frame containing a preamble comprising a predetermined number of preamble symbols to form a known preamble codeword;

(b) cross-correlating 84 successive pairs of symbols in the received frame to generate cross-correlation outputs;

(c) converting the cross-correlation output into a binary sequence;

(d) selecting a first number of consecutive bits from the binary sequence, wherein the selected number matches the number of preamble symbols;

(e) selecting a codeword from a plurality of predetermined codewords that most closely matches the selected number of bits;

(f) comparing 90 the selected codeword to the preamble codeword;

(g) in case the selected codeword sufficiently matches the preamble codeword, deciding 92 that synchronization has been obtained; and

(h) in case the selected codeword does not sufficiently match the preamble codeword, performing the following steps:

(i) acquiring an additional symbol;

(j) cross-correlating the additional symbol with a previously received symbol (immediate preceding received symbol) to generate a cross-correlation value;

(k) converting the cross-correlation value into a first binary value;

(l) Concatenating the first binary values into the binary sequence;

(m) deleting the opposite binary values from the binary sequence; and

(n) repeating steps (e) through (m) until a predetermined stopping criterion (e.g., synchronization) is reached.

It should of course be appreciated that a soft decoder can also be used to analyze the cross-correlation values. In particular, the soft decoder acts directly on the cross-correlation values (to produce decoded codewords), without the need to convert the cross-correlation values to binary sequences.

Although in the preferred embodiment, it is described with reference to a preamble being placed at the beginning of an OFDM frame, it should be appreciated that the codewords in the preferred embodiment can also be placed at the end of OFDM. In particular, the relative placement of the preamble is not important as long as the relative placement is established in advance and is not changed.

The description of the preferred embodiments focuses on the use of cyclic codes as they are easier to implement as there is no need to find out the specific code that the preamble has been embedded with. In particular, if another code is detected, which is known to have, for example, two steps (or loop iterations) from the desired code, then there is no need to perform further DFT and cross-correlation operations, since the correlation starting point of the OFDM frame is equally two locations away. It should be understood, however, that the preferred embodiments are not limited to the use of cyclic codes. In particular, the preferred embodiment can employ any other binary error correction code. However, it is appreciated that other codes may have greater correction capabilities, although these other codes typically have more complex implementations.

Substitutions and modifications may be made to the above-described embodiments without departing from the scope of the invention.

Claims (5)

1. An OFDM frame synchronization method, characterized by comprising the steps of:

(a) receiving a transmitted OFDM frame containing a preamble, the preamble comprising a predetermined number of preamble symbols, the preamble symbols to form a known preamble codeword;

(b) cross-correlating (84) successive pairs of symbols in the received frame to generate a cross-correlation output;

(c) selecting a codeword from a plurality of predetermined codewords, wherein the selected codeword most closely matches a number of cross-correlation values selected from the cross-correlation output;

(d) comparing (90) the selected codeword with the preamble codeword;

(e) -deciding (92) that synchronization has been obtained in case the selected codeword matches the preamble codeword; and

(f) in case the selected codeword does not match the preamble codeword, performing the following steps:

-obtaining additional symbols;

-cross-correlating the additional symbol with a previously received symbol to generate a further cross-correlation value;

-connecting the further cross-correlation value to the cross-correlation output;

-removing the opposite cross-correlation value from the cross-correlation output; and

-repeating steps (c) to (f) until a predetermined stop criterion is reached.

2. The OFDM frame synchronization method of claim 1, wherein the step of selecting the codeword from a plurality of predetermined codewords comprises the steps of:

-converting the cross-correlation output into a binary sequence;

-selecting a first number of consecutive bits from the binary sequence, wherein the selected number matches the number of preamble symbols; and

-selecting from a plurality of predetermined code words, the code word that most closely matches the number of selected bits; and

the steps relating to connecting the further cross-correlation values to the cross-correlation output and relating to deleting the opposite cross-correlation values from the cross-correlation output comprise the steps of:

-converting the cross-correlation value into a first binary value;

-connecting the first binary value to the binary sequence; and

-deleting the opposite binary values from the binary sequence.

3. The OFDM frame synchronization method of claim 1 or 2, wherein before the step of receiving the OFDM frame including the preamble, comprising the steps of:

-providing a first plurality of short symbols having the same sign;

-providing a second plurality of short symbols, the signs of the short symbols being arranged such that one or more cross-correlations of one or more pairs of each successive short symbol generates a pre-selected error correction codeword;

-concatenating the first and second plurality of short symbols to form a preamble; and

-providing the OFDM frame with the preamble before transmission.

4. The OFDM frame synchronization method of claim 3, wherein the step of providing the second plurality of short symbols comprises the steps of:

(1) providing a third plurality of short symbols having the same sign;

(2) selecting a binary error correction code having the same number of bits as the number of short symbols in the third plurality;

(3) using a first short symbol in the third plurality to form the first short symbol in the second plurality;

(4) setting a sign of a first short symbol in the second plurality to match a sign of a first bit in the selected codeword;

(5) using a next short symbol in the third plurality to form a next short symbol in the second plurality;

(6) setting the sign of the next short symbol in the second plurality to match the sign of the multiplication of the corresponding bit in the selected codeword with the sign of the previous short symbol in the second plurality; and

(7) repeating steps (5) and (6) until the second plurality of short symbols is completed.

5. The OFDM frame synchronization method of claim 1 wherein the step of selecting from a plurality of predetermined codewords comprises the step of selecting from a plurality of predetermined cyclic codewords.