Multi user detection using equalization and successive interference cancellation

Multi-user detection using equalization and continuous interference cancellation

Technical field

The present invention generally relates to wireless communication systems. In particular, the present invention relates to the detection of multi-user signals in a wireless communication system.

BACKGROUND OF THE INVENTION A typical wireless communication system includes base stations that communicate with wireless transmit/receive units (WTRUs). Each base station has a related operating area, which communicates with WTRUs in this operating area. In some communication systems, such as coded multi-directional proximity (CDMA) systems, the same frequency spectrum is used to transmit multiple communications. Such communications are usually distinguished by coding.

Since multiple communications can be transmitted in the same frequency spectrum and at the same time, a receiver in this system must distinguish these multiple communications. One solution for detecting these signals is matched filtering. In the matched filtering method, a communication transmitted with a single code is detected. Other communications are treated as interference. To detect multiple codes, a corresponding number of matched filters must be used. These signal detectors have low complexity, but suffer from multi-directional proximity interference (MAI) and inter-symbol interference (ISI).

Other signal detectors try to eliminate interference and ISI from other users, such as parallel interference cancellers (PICs) and continuous interference cancellers (SICs). Such receivers tend to have better performance, at the expense of increased complexity. Other signal detectors jointly detect multiple communications, which is called joint detection. Some joint detectors use the Cholesky decomposition method for a minimum mean square error (MMSE) detection and forced zero block equalizers (ZF-BLEs). Such detectors tend to have better performance, but high complexity.

Therefore, it is hoped that there can be other multi-user detection solutions.

Summary of the invention

Receive multiple signals in a shared spectrum. A received vector is generated from the samples of the received user signal. The received vector is divided into multiple sections. For each segment, continuously determine each user or signal group (signal group with similar received power) by determining a user/signal group and removing the component of the user/signal group from the received vector symbol. For example, the symbol of each user/signal group is determined by means of equalizing the channel and then despreading. Combine the determined symbols corresponding to each segment into a data vector.

Description of the drawings

Figure 1 is a simplified diagram of an equalized continuous interference cancellation (EQ-SIC) receiver.

Fig. 2 is a diagram of a preferred segmentation method for receiving vector r.

Figure 3 is a schematic diagram of an EQ-SIC device.

Figure 4 is a flowchart of an EQ-SIC receiver.

detailed description

The preferred application of the preferred embodiment lies in the one-frequency division duplex (FDD) mode of the third-generation partnership project (3GPP) wide-band coded multi-directional proximity (W-CDMA) communication system. However, the preferred embodiment can also be applied to many wireless communication systems.

These preferred embodiments can be used in a wireless transmit/receive unit (WTRU) or a base station. A WTRU is not limited to include a user equipment, a mobile station, a fixed or mobile subscriber unit, a pager, or any other device type capable of operating in a wireless environment. A "base station" non-limitingly includes a base station, a B-level node (Node-B), a network point controller, an access point, or other interface devices in a wireless environment. In addition, these preferred embodiments may be applied to WTRUs that communicate with each other.

Figure 1 is a schematic diagram of a preferred equalization/continuous interference cancellation (EQ-SIC) receiver. Preferably, most of the components shown in FIG. 1 except for the antenna 20 are implemented as a single integrated circuit. Alternatively, the individual components may be discrete components or a mixture of integrated circuits and/or discrete components.

Multiple communications are received by an antenna 20 or antenna array of the receiver. A sampling device 22 (such as single or multiple analog-to-digital converters (ADCs)) samples the received signal to generate a received vector r.

The received vector is processed by a segmentation device 24 to generate sections r1,...rs of the received vector r. Figure 2 is a diagram of a better segmented architecture, but other architectures are also possible. As shown in Fig. 2, the received vector r is divided into a plurality of sections r1,...rs. Preferably, these sections partially overlap as shown in the figure. The amount of overlap is preferably twice the length of the impulse response minus one chip, that is, 2×(W-1). Even if the length of the section is limited, this overlap helps equalize all the chips. In the case of a known section, all the chips constituting the part of the section are equalized. For example, the part of r2 is bounded by a dotted line. The last chip in this section will extend to the next section by W-1 chips. Conversely, the farthest chip before the first chip extending into the area is W-1 chips before the first chip. Therefore, all the chips constituting the part and the chips not in the part can be equalized, effectively removing the components of the chips in the part.

Although the amount of overlap shown in the figure is roughly twice the impulse response, a larger amount of overlap can also be used. Depending on the actual receiver application, a larger amount of overlap may be advantageous. In one embodiment, the EQ-BIC device may use a prime factor algorithm (PFA) fast Fourier transform (FFT) type application. The amount of overlap can be extended to reach a desired optimal PFA or FFT length. In other applications, the best non-overlapping part can vary depending on the signal being processed. For example, in the time division duplex (TDD) mode of 3GPP W-CDMA, the length of the data field may vary depending on the burst type. Therefore, the optimal segment length of one burst data block may not be the best case for another burst data block. When a consistent hardware configuration is to be used, a predetermined segment size can be used. Different amounts of overlap can be used to promote different burst data block lengths.

A channel evaluation device 26 evaluates the channel response of each received user signal. Generally speaking, the channel response is evaluated using a reference signal (such as a amble or a midamble sequence), but other techniques can also be used. The estimated channel response is represented by a channel response matrix H in FIG. 1.

FIG. 3 is a diagram of a preferred EQ-SIC device 28. In an application, all user signals are classified, for example, the received power of the signals is used for classification. For the user with the highest received power, the received vector ri is equalized by an equalizer 341 using the channel response related to the user (user 1) to generate a spread-spectrum data vector si1. A despreader 361 generates the soft symbols of the user data using the code used for the user signal. A hard decision device 381 performs a hard decision on the user's soft symbol to generate a hard symbol vector di1. The measured hard symbol is used to determine the component ril of the spread-spectrum data vector by the user 1. A subtractor 421 subtracts the user 1 component from the section to generate a new section xi1 from which the user 1 component has been removed. A second user (User 2) with a second highest received power level is processed in a similar manner. The hard symbol di2 of the user 2 is measured by an equalizer 342, a despreader 362, and a hard decision device 382. An interference construction device 402 and a subtractor 422 are used to remove the user 2 component of Xi1. This procedure is repeated until the last user K. For the Kth user, only the hard symbol diK is determined by an equalizer 34K, a despreader 36K, and a hard decision device 38K.

If the EQ-SIC receiver is used in a base station, it will generally restore the hard symbols from all user signals. However, in a WTRU, the WTRU EQ-SIC receiver may only have the signal of one target user. Therefore, the continuous processing for each user can be stopped after the hard symbol of the target user's signal has been restored.

Although the above is to detect the signal of each user independently, it is also possible to jointly restore the signals of multiple users. In this application, the received signal power is used to group users into groups. Continuous processing will be performed for each group in sequence. For example, the first group of data is detected and then the first group of data is eliminated from the receiving section, and then the second group is performed.

After the data of each user in a section has been measured, a section storage device 30 stores the data vector, such as di. In order to reduce the storage size, it is better to cut the section to remove the non-target part and leave only the target section part. A sector recombination device 32 generates a data vector d with data from all sectors. The vector is usually formed by serially combining the data of each user in each sector. For example, the data d11 of user 1 from section 1 and the data d12 of user 1 from section 2 are serially combined.

Figure 4 is a flow chart of an EQ-SIC receiver. At the beginning, a reception vector r is generated, step 50. Perform a channel assessment for all users, step 52. The received vector is segmented into r1,...rn, step 54. Make each section processed, step 56. Determine a user with the highest received power for an i-th segment, step 58. Equalize the received vector with the user, step 60. Use the user's code to de-spread the obtained spread vector, step 62. A hard decision is made on the despreading data, step 64. Determine the user's component of the received vector, step 66, subtract the user's component from the received vector, step 68. Using the reduced code reception vector as the reception vector in the subsequent steps, repeat steps 60-68 and step 70 for the user with the second highest received power. Store the result of the section and repeat steps 58-70, step 72 for each remaining section. Combine the stored sections into a data vector d, step 74. The rate at which the channel evaluation is performed or updated may vary from application to application, because the update rate depends on the nature of the time variation of the wireless channel.

Preferably, the equalization operation of each stage of the EQ-SIC device 28 is performed by FFT, but other methods of execution may also be used. One possible way of implementation is as follows. Each receiving section can be regarded as a signal pattern according to Equation 1.

ri=Hs+n Equation 1H is the channel response matrix. n is the noise vector. s is the spreading data vector, which is the convolution of the spreading code C of the user or group and the data vector d of the user or group, as shown in Equation 2.

s=Cd Equation 2 is used to solve the two schemes of Equation 3 first with an equalization stage and then with a solution expansion stage. Equalize each received vector section ri, step 54. The first equalization scheme uses a minimum mean square error (MMSE) solution. The MMSE solution of each extended section is based on Equation 4A.

i=(HsHHs+σ2Is)-1HsHri equation 4Aσ2 is the noise variance, and Is is the identity matrix of the extended matrix. (.) H is a complex conjugate transpose operation or Hermetian operation. The forced return to zero (ZF) solution is based on Equation 4B.

i=(HsHHs)-1HsHri Equation 4B Another alternative, write Equation 4A or 4B as Equation 5.

i=Rs-1HsHri Equation 5 corresponds to the MMSE Rs system is defined in accordance with Equation 6A.

Rs=HsHHs+σ2Is Equation 6A Another alternative, the Rs of ZF is based on Equation 6B.

Rs = HsHHs solving Equation 6B party program 5 a preferred embodiment of the system using a fast Fourier transform (an FFT), as in equation 7 and 8, a further embodiment the solution for Equation 5 Cholesky decomposition.

Rs=Dz-1Dz=(1/P)Dz×Dz equation 7Rs-1=Dz-1-1Dz=(1/P)Dz××Dz equation 8Dz is the Z-point FFT matrix and is right The angular matrix, which has an FFT diagonal of the first row of an approximate round determinant of the Rs matrix. The approximate wheel ring determinant can be expressed in any row of the Rs matrix. It is better to use a complete row with the most elements.

In the frequency domain, the FFT solution is based on Equation 9.

F(s~^)=Σm=1MF(h~m)*F(r~m)F(q~)]]>where F(x~)=Σn=0P-1x(n )e-j2πknN,]]> where k=0,1,..., P-1 equation 9 is the Kronecker product. M is the sampling rate. M=1 is the chip rate sampling and M=2 is the double chip rate sampling.

After the Fourier transform F() of the spread data vector has been determined, the spread data vector is determined by adopting an inverse Fourier transform.