JPWO2006030479A1 - Turbo coded bit allocation system and turbo coded bit allocation retransmission method in MIMO-OFDM system - Google Patents

Abstract

Turbo coded bit allocation system in MIMO-OFDM system. In this system, the V-BLAST processing unit (110) calculates the norm of each row of the pseudo inverse of the channel matrix for each received signal vector during the V-BLAST detection. Feedback vector information indicating the antenna position of the first layer, which is the row having the minimum norm, is stored in the vector information output unit (116) of the receiver. The vector information output unit (116) of the receiver transmits the feedback vector information to the transmitter at a desired cycle via the error-free channel (118). The bit allocation unit (106) allocates systematic bits and parity bits to different antennas based on the feedback vector information. The bit rearrangement unit (114) rearranges each bit to the original position in the turbo code.

Description

  The present invention relates to a turbo coded bit allocation system and a turbo coded bit allocation retransmission method in a multiple-input multiple-output (MIMO) communication system using orthogonal wave frequency division multiplexing (OFDM).

Simultaneous transmission of multiple data streams is implemented in a MIMO communication system using multiple (N _T ) transmit antennas and multiple (N _R ) receive antennas. Depending on the usage, the MIMO system can contribute to performance improvement by spatial diversity or increase system capacity by spatial multiplexing. This improvement is possible because the wireless communication system has random fading and multipath delay spread.

  The plurality of communication channels existing between the transmitting antenna and the receiving antenna usually have different link states while changing with time. A MIMO system with feedback provides channel state information (CSI) to the transmitter and allows the use of methods such as link adaptation and water filling to provide a higher level of performance.

  A well-known technique for increasing system capacity gain through spatial multiplexing is the BLAST technique. This technique is discussed in [1].

  The MIMO technique was originally designed for narrowband wireless systems, ie flat fading channels. Therefore, it is difficult to obtain a high effect in the frequency selective channel. Thus, OFDM is used in conjunction with a MIMO system to overcome the frequency selective channel proposed in the wireless environment.

  Using Inverse Fast Fourier Transform (IFFT), OFDM can transform a frequency selective channel into a set of independent, parallel frequency constant subchannels. Since the frequencies of these subchannels are orthogonal and overlap each other, spectrum efficiency is improved and intercarrier interference is minimized. The multipath effect is further reduced by adding a cyclic prefix to the OFDM symbol.

  Adaptive bit loading in OFDM systems has been discussed in various technical papers. By changing the number of bits allocated to OFDM subcarriers, adaptive bit loading aims to optimize the data rate without degrading system quality. This technique works on the fact that each different subcarrier has a variable attenuation depending on the channel conditions. Allocation decisions are typically made with specific feedback information such as channel state information (CSI) and signal-to-noise ratio (SNR) for each subcarrier.

  Many conventional communication systems use turbo coding as an error correction method. The reason is that characteristics close to the Shannon limit can be obtained. The problem of unequal power allocation to systematic and parity bits has been discussed. In Non-Patent Document 2, when the systematic bit protection is lower than the parity check bit, a better performance is obtained as the interleaver length increases, and this effect becomes stronger as the block length increases. It is shown.

  On the other hand, Non-Patent Document 3 proposes that a large power should be allocated to systematic bits when the SNR is very low.

"V-BLAST: an architecture for realising very high data rates over the rich-scattering wireless channel" by P W Wolniansky et al in the published papers of the 1998 URSI International Symposium on Signals, Systems and Electronics, Pisa, Italy, Sep. 29 to Oct. 2, 1998. “Unequal power allocation to the turbo-encoder output bits with application to CDMA systems” by Atousa H. et al. S. Mohammadand Amir K.M. Khandani in the transaction letter of IEEE Transactions on Communication, Vol. 47, no. 11, Nov 1999. “Optimizing the energy of differential bits stream of Turbo codes” by J. et al. Hokfeld and T.W. Maseng in Proc. Turbo Coding Seminar, Lund, Sweden, Aug. 1996, pp. 59-63.

  However, until now, no technique related to a method for assigning turbo coded bits in a MIMO-OFDM system has been disclosed.

  An object of the present invention is to provide a turbo encoded bit allocation system and a turbo encoded bit allocation retransmission method capable of performing optimal turbo code bit allocation in a MIMO-OFDM system. Another object of the present invention is to achieve improved system performance through information-based information bit allocation without overly complicating the system.

  The present invention assigns systematic bits and parity bits to OFDM subcarriers transmitted by different transmission antennas based on feedback information provided by the V-BLAST processing unit on the receiver side.

  According to the present invention, optimal turbo code assignment can be performed without excessively complicating the system, and improvement in system performance can be achieved.

Block diagram showing an overview of the system of the present invention Block diagram of transmitter of MIMO-OFDM communication system Block diagram of receiver of MIMO-OFDM communication system The flowchart which shows the V-BLAST signal processing method and information acquisition in one Example of this invention Diagram showing the detection processing performance of each transmission layer on the assumption that error propagation does not occur

  FIG. 1 is a diagram showing a system of the present invention including a bit allocation unit 106 on the transmitter side and a bit rearrangement unit 114 on the receiver side.

  The V-BLAST processing unit 110 performs V-BLAST processing that separates the received signal into data streams corresponding to the multiple antennas of the transmitter. The V-BLAST processing unit 110 calculates the norm in each row of the pseudo inverse of the channel matrix for each received signal vector during the V-BLAST detection. Since the row having the minimum norm has the highest post-detection SNR, the first detection target layer (hereinafter referred to as “first layer”) is used in order to obtain better performance of the entire system. Information on the feedback vector indicating the antenna position of the first layer is stored in the vector information output unit 116 of the receiver.

  The vector information output unit 116 of the receiver transmits the feedback vector information to the transmitter at a desired cycle via the error-free channel 118.

  The input data is turbo encoded by the turbo encoding unit 102. The systematic bits and parity bits generated by turbo coding are separated by the bitstream separation unit 104. The bit allocation unit 106 allocates systematic bits and parity bits to different antennas based on the feedback vector information.

  On the other hand, the bit rearrangement unit 114 rearranges each bit demapped by the demapping unit 112 to the original position in the turbo code.

Here, in the V-BLAST detection, the diversity level increases because the layer that is detected later has more signals that are erased in the previous stage. First layer having the highest post-detection SNR is to be detected first and foremost in the V-BLAST detection, it is the lowest (the slope of BER characteristics of the horizontal axis and the _E b / _{N o)} diversity level, the highest Error rate.

  In one embodiment of the present invention, parity bits are better protected than systematic bits by assigning systematic bits to antennas that are likely to have the highest post-detection SNR. In this allocation method, the larger the interleaver length, the better the performance is expected.

  In another embodiment of the invention, systematic bits are better protected than parity bits by assigning parity bits to the antenna with the highest post-detection SNR in the previous transmission. This allocation method helps improve performance when the SNR is very low (eg, below a predetermined threshold).

  In one variation of the embodiment of the present invention, the antenna rank according to the post-detection SNR obtained from V-BLAST is transmitted to the transmitter. Adaptive bit loading can be performed by allocating a larger number of bits to subcarriers transmitted by a lower SNR antenna in the bit allocation unit 106. This is because a higher-order modulation method can be used if a higher diversity effect is obtained during detection.

  In another variation of the embodiment of the present invention, adaptive power allocation is performed by the power allocation unit 108 based on long-term statistics obtained from feedback information. An antenna having a larger number of times of becoming the first layer is assigned a larger power. It has been shown that such an allocation results in improved performance.

  The above embodiments aim to improve system performance with a low level of complexity. The present invention utilizes feedback information obtained from the V-BLAST technique already provided in the system. In addition, adaptive allocation and bit loading are performed based on the SNR ranking between antennas. Unlike conventional methods that require SNR to be intensively calculated for each symbol, the present invention simplifies the overall processing procedure, thereby reducing the complexity of the processing mechanism and allowing feedback with shorter delays. To.

FIG. 2 is a block diagram of a transmitter 200 in a multiple-input multiple-output communication system (ie, MIMO-OFDM system) that utilizes orthogonal frequency division multiplexing. FIG. 3 is a block diagram of the receiver 300 of the system. Both figures show a system using two transmit antennas and two receive antennas, but the present invention can be extended to a system using multiple (N _T ) transmit antennas and multiple (N _R ) receive antennas. is there. In the following description, a 1/2 turbo coding rate is used as an example.

  In the transmitter 200, the cyclic redundancy check (CRC) code is first added to the input data by the CRC adding unit 202. Thereafter, turbo encoding is executed by the turbo encoding unit 204. At the output of the turbo encoder 204, the systematic bits and the parity bits are divided into the same individual streams 206 and 208, respectively. Each stream of encoded data undergoes interleaving separately at interleaver 210 to reduce burst errors in the data. The separated systematic bits and parity bits are then input to the bit allocation unit 212. Based on the information provided from the receiver, the bit allocation unit 212 allocates encoded bits to each allocated transmission antenna. Details of the allocation procedure will be described later. Next, multi-level modulation constellation symbol mapping is performed in the mapping unit 214 for the data directed to each transmission antenna. A pilot signal is inserted into the mapped signal by pilot insertion section 216. Channel estimation at the receiver is facilitated by inserting pilot signals.

  Before performing OFDM modulation, the S / P converter 218 converts a serial data stream into a parallel data stream. IFFT section 220 makes generated subcarriers orthogonal to each other. After the parallel data is converted into serial data by the P / S converter 222, a cyclic prefix (cyclic prefix) for reducing the multipath effect is added to the OFDM symbol by the CP adder 224. Before transmission, the digital signal is converted into an analog signal by the D / A converter 226. After passing through various processes in each transmitter chain, the signal is ready for transmission by its assigned transmit antenna 228.

  In the receiver 300, processing reverse to the above is performed on the received signal from the receiving antenna 302, that is, analog to digital conversion (A / D conversion unit 304), cyclic prefix removal (CP removal unit 306), Processing such as serial / parallel conversion (S / P conversion unit 308), fast Fourier transform (FFT unit 310), and parallel / serial conversion (P / S conversion unit 312) is performed. Since the received signal consists of overlapping signals from a plurality of transmission antennas, it is necessary to separate the signals into individual streams. In this case, a V-BLAST decoder 314 that utilizes techniques such as zero forcing (ZF) or minimum mean square error (MMSE) is used to perform this function.

  After demapping is performed by the demapping unit 316, each demapped bit is rearranged at its original position. Such rearrangement is performed according to an assignment pattern stored in the receiver. This is the same pattern used for bit allocation on the transmitter side. The output from the bit rearrangement unit 318 is composed of a separate stream of systematic bits and parity bits. Further, deinterleaving (deinterleaver 320) and decoding (decoding unit 322) are performed. Subsequently, a cyclic redundancy check (CRC processing unit 324) is executed for each frame to confirm that the data is correct. If the inspected frame is determined to be error free, an acknowledgment (ACK) is sent to the transmitter and the transmitter does not retransmit the packet. If there is an error, a negative acknowledgment (NACK) is sent to the transmitter 200 to request retransmission.

  FIG. 4 is a flowchart showing a V-BLAST signal processing method and information acquisition according to an embodiment of the present invention. Based on the information obtained from the V-BLAST process, the SNR ranking of the antennas is performed. The V-BLAST technique using the ZF criterion and the antenna ranking procedure will be described below.

  The signal received by each receiving antenna consists of a mixture of signals from each transmitting antenna. Therefore, V-BLAST aims to detect hybrids and separate them into valid data streams. The V-BLAST technique is performed on each column of symbols obtained from all receive antennas). A channel matrix obtained from channel evaluation corresponding to each received vector is also required for the V-BLAST processing.

The first step after setting the received vector (r _i ) and the corresponding channel matrix (H _i ) set is to calculate the pseudo inverse of H _i to obtain the set of matrices G _i (step 404).

After obtaining the pseudo inverse of the channel matrix, the norm of each row of G _i is calculated (step 406). The purpose of this step is to select the first layer. It has already been proven that better performance can be obtained if detection is first performed on the layer that exhibits the highest post-detection signal-to-noise ratio (SNR). This corresponds to detecting the line of G _i having the smallest norm.

Therefore, the row of G _i having the minimum norm is selected as the _zeroization vector (w _k ) (step 408). This vector, except for signals transmitted in k ^th, zeroing all signals (disabled). The V-BLAST processing unit 110 detects a symbol transmitted from the k ^th transmission antenna by internally generating a reception vector (r _i ) and a weight vector (w _k ) (step 410).

  The detected symbol is regenerated by slicing to the nearest value in the signal constellation (step 412). By doing so, the evaluation for signal erasure executed in the next step is efficiently processed.

  If one layer is detected, subsequent layer detection processing can be performed efficiently. By subtracting the detected signal portion from the vector of the received signal (step 414), the number of layers to be zeroed in the next step is reduced. This erasure process corrects the received signal vector to a vector with a reduced interference signal component. It also reduces detection complexity.

By clearing the k ^th column of H _i corresponds to (step 416), it is necessary to also modify the channel matrix.

  The entire process is repeated until all layers are successfully detected (step 418). Note that the V-BLAST detection is continued for other sets of received vectors.

  Since the norm is an indication of the SNR after detection of each transmission layer, the present invention uses this information for turbo coded bit allocation at the transmitter. For each symbol in the frame, the receiver stores the antenna position of the first layer having the minimum norm (maximum SNR) as a vector. Only the first layer need be saved, and subsequent layers need not be saved. The receiver then feeds back the information vector to the transmitter via the error-free channel. The transmitter uses this vector for bit allocation of the next transmission frame. This storage and feedback process is repeated at the desired period. These two steps may be performed every time a frame is received at the receiver, or may be performed every desired number of frames. As a general guideline, feedback is performed in a shorter cycle for a radio channel that fluctuates quickly. On the other hand, slow-changing channels attempt to update information at longer intervals to save resources and keep complexity at a minimum level.

FIG. 5 is a diagram showing the performance of each detected transmission layer on the assumption that no error propagation occurs during V-BLAST detection. In FIG. 5, the horizontal axis represents E _b / N 2 _O and the vertical axis represents BER (Bit Error Rate). In FIG. 5, communication is performed between a transmitter of four transmitting antennas and a receiver of four receiving antennas, and the signal from the transmitting antenna 1 (Tx1) is sequentially separated to the signal of the transmitting antenna 4 (Tx4). An error curve is shown.

  Theoretically, the diversity level is the lowest when the first layer is detected. For each detected layer, each receive antenna is much constant, but the detected layer is erased, resulting in an increase in system diversity level as the layer progresses. This is clearly shown in FIG. The error curve of antenna 1 detected in the first decoding step gradually falls as the SNR increases (diversity level of antenna 1). The error curves of antennas 2, 3, 4 fall much more rapidly than that of antenna 1. By taking advantage of subtracting previously detected symbols, the diversity level is increased to 4 in subsequent layers. Level 4 occurs when only one signal remains to be detected by the four receiving antennas. Therefore, it can be concluded from this observation that the first layer after ranking has the highest post-detection SNR, but its performance is lower than other transmission layers detected after that. is there.

  Previous studies have shown that system performance changes when systematic bits obtained from turbo coding and parity bits have different protection levels. How to protect these bits to gain performance depends on system parameters and channel conditions.

  Therefore, based on the above-described research results, the present invention aims at obtaining an optimal allocation of systematic bits and parity bits to the allocated transmission antennas.

  As described above, the receiver sends feedback vector information including the first layer antenna position detected for each sequence of processed received symbols to the transmitter.

  In one embodiment of the present invention, the transmitter bit allocation module allocates systematic bits to the antenna specified by the feedback vector. That is, systematic bits are assigned to the antenna that is highly likely to have the highest post-detection SNR, that is, the first layer. This means that systematic bits have a higher error probability than parity bits detected in subsequent layers. This is consistent with the research results that when the systematic bit protection is lower than the parity check bit, better performance can be obtained by increasing the interleaver length. Therefore, by using this allocation method, better protection of the parity bits is realized. The larger the interleaver length, the better the performance.

  In other embodiments of the invention, systematic bits are better protected than parity bits. This is in contrast to the previous examples. In this embodiment, rather than assigning systematic bits to the first layer, parity bits are transmitted at this layer. This allocation method is used when the SNR is very low, i.e. in severe channel conditions.

  In one variation of the above embodiment, the receiver feeds back not only the antenna position of the first layer but also the antenna position of the subsequent layer. Therefore, the rank between the antennas by the SNR after detection of each antenna is obtained at the transmitter. Based on this ranking, some form of adaptive bit loading can be performed. For example, a larger number of bits may be allocated to subcarriers transmitted by the antenna having the lowest SNR order. Since this antenna will probably have the highest diversity when V-BLAST is detected, a predetermined error rate can be maintained even with higher order modulation schemes.

  In another variation of the above embodiment, adaptive power allocation can be performed based on long-term statistics of feedback information. Since the feedback information includes the antenna position of the first layer for each column of received symbols, the number of times each antenna becomes the first layer can be calculated. In order to enhance system performance, it has been shown that transmitters are more likely to allocate higher power to earlier detection stages. Therefore, the overall error rate can be reduced by giving larger power to the antenna having a higher number of times of becoming the first layer in the long-term statistics.

  Although the above description is considered as a preferred embodiment of the present invention, the present invention is not limited to the disclosed embodiment, and can be realized in various forms and embodiments, and is within the scope of the present invention. Is to be determined with reference to the following claims and their equivalents.

  The present invention is suitable for use in a multiple-input multiple-output (MIMO) communication system using orthogonal wave frequency division multiplexing (OFDM).

Claims (8)

A turbo coded bit allocation system in a multiple input multiple output (MIMO) communication system using orthogonal wave frequency division multiplexing (OFDM) comprising:
A V-BLAST signal processing unit that performs V-BLAST processing for separating a received signal into a data stream corresponding to a plurality of antennas of the transmitter, and feedback vector information obtained by the V-BLAST processing to the transmitter A vector information output unit for sending,
A turbo encoder that generates systematic bits and parity bits by turbo encoding, and a bit that allocates a separate stream of the systematic bits and the parity bits to individual antennas based on the feedback vector information And an assigning unit. The system of claim 1 comprises:
The transmitter further includes a separate interleaver corresponding to each stream of the systematic bits and the parity bits and preventing the systematic bits and the parity bits from being mixed before the bits are allocated. The system of claim 1 comprises:
The receiver further includes a bit rearrangement unit that rearranges the demapped bits to their original positions based on the allocation pattern at the previous transmission. A turbo coded bit allocation method in a multiple input multiple output (MIMO) communication system using orthogonal frequency division multiplexing (OFDM) comprising:
In the receiver, performing a V-BLAST process for separating a received signal into a data stream corresponding to the multiple antennas of the transmitter, and sending feedback vector information obtained by the V-BLAST process to the transmitter. Equipped,
In the transmitter, generating systematic bits and parity bits by turbo encoding, and assigning separated streams of the systematic bits and the parity bits to individual antennas based on the feedback vector information. It has. The method of claim 4 comprises:
In the receiver, information indicating the antenna position having the highest post-detection SNR as the feedback vector information is sent to the transmitter;
In the transmitter, systematic bits are allocated to the antenna of the feedback vector information. The method of claim 4 comprises:
In the receiver, information indicating the antenna position having the highest post-detection SNR as the feedback vector information is sent to the transmitter;
When the SNR is lower than a predetermined threshold, the transmitter assigns a parity bit to the antenna of the feedback vector information. The method of claim 4 comprises:
In the receiver, as the feedback vector information, information indicating the SNR rank of the antenna is sent to the transmitter,
In the transmitter, a larger number of bits are allocated to subcarriers transmitted by an antenna having a lower SNR order, while a smaller number of bits are allocated to subcarriers transmitted by an antenna having a higher SNR order. The method of claim 4 comprises:
In the transmitter, the number of times each antenna has the highest post-detection SNR is calculated, and the higher the number of antennas, the greater the power.

Images