High-speed cell search system for CDMA

Description: TECHNICAL FIELD [Detailed description of the invention]

【0001】

[Technical field to which the invention belongs]

The present invention relates to a high-speed cell search system for CDMA, and relates to a high-speed search circuit used in a CDMA communication system in which an interference wave component power is larger than that of a target signal component power. More specifically, the present invention covers a plurality of signals. By performing synchronous addition, the S / N is increased to perform a search, and long codes are identified and long codes are narrowed down using orthogonal codes.

【0002】

[Conventional technology]

In CDMA cellular systems, mobile stations need to establish diffusion code sequence synchronization before establishing carrier or clock synchronization. It can take a long time to establish this code synchronization, and shortening it has become an important issue for CDMA.

【0003】

[Problems to be Solved by the Invention]

The present invention has been devised to solve the following two problems.

【0004】

The mobile station performs a cell search for detecting the connected cell on the assumption of initial synchronization of the diffusion code.

【0005】

Conventionally, in the inter-base station synchronization system represented by IS-95, each base station flows synchronously with a certain fixed offset amount (a value that determines where to start from the diffusion code), and each base station flows. It is possible to use a common spreading code, and since the mobile station only needs to search for one type of spreading code, the search time can be shortened.

【0006】

However, to use this method, it was necessary to synchronize between the base stations, which required an external system.

【0007】

Therefore, an asynchronous system between base stations that does not require an external system is desired. However, since the base stations are asynchronous, one spreading code cannot be shared by each base station. Therefore, different types of spreading codes are used by each base station, but there is a drawback that a search for all types of spreading codes is required in the process of initial synchronization, resulting in an increase in search time.

【0008】

In addition, the data quality of the information convolutionally encoded by the coding gain of the Viterbi processor is good, but the Viterbi processor does not work because the data is not read when the synchronization of the spread code sequence is established, and there is no coding gain. Will be detected by. If the synchronization establishment is not completed, no carrier synchronization will start, so it is necessary to detect with good quality even in the absence of this coding gain. That is, if there is a probability of erroneous detection, there is a drawback that the search is repeated and the search time is increased.

【0009】

The present invention relates to a search method used for CDMA in which the interference wave power is larger than the above-mentioned target signal power, and more specifically, a high-speed search is performed using a long code mask symbol spread only by a short code. It is related to the cell search of the long code mask method to be performed.

【0010】

Here, the long code mask method is a configuration in which a long code spread is masked at regular intervals in an asynchronous cellular system between CDMA base stations and has a symbol (mask symbol) spread only by a short code, and the mobile station has this mask symbol. It is a method to realize high-speed cell search by using (described in RCS96-74, a credential technical report of the Wireless Communication System Study Group of the Electronic Communication Information Society).

【0011】

The long code mask method that has been announced so far is to detect the timing of the long code by using the mask symbol, and then identify the long code by using a normal sliding correlator. Therefore, for example, when the number of accommodated users increases and the interference wave power increases, the probability of erroneous detection increases, the search is repeated, and the search time increases. Further, if the number of long codes to be identified is large, there is a drawback that the search time is increased accordingly.

【0012】

On the other hand, the present invention makes it possible to identify or narrow down long codes by repeating M long code mask symbols L times, and further improves SNR by performing in-phase addition over M symbols. A high-speed cell search is realized by identifying or narrowing down long codes at an early stage while reducing the error probability.

【0013】

Furthermore, the present invention is configured by a Walsh-Hadamard code composed of M symbols in a CDMA system that realizes an asynchronous system between base stations by spreading each base station with a code called a long code peculiar to the base station. High-speed cell search is realized because the data includes information for identifying the base station, and the long code can be narrowed down or the long code can be specified based on the information.

【0014】

Further, the present invention can be applied to orthogonal demodulation having I system and Q system, and power addition is performed after in-phase addition to enable higher-speed cell search.

【0015】

Further, the present invention can be realized with a small circuit scale by the ring buffer that adds with the polarity corresponding to the orthogonal code, so that a configuration suitable for low power consumption is possible.

【0016】

Further, in the present invention, since the barker sequence can be used for specifying the frame position, the data composed of the orthogonal codes will not be erroneously detected as the frame synchronization signal.

【0017】

Furthermore, the present invention can be applied to serial search, parallel search, or serial / parallel mixed search, and the degree of parallelism, that is, the search detection time can be selected according to the feasible circuit scale, and the design is flexible. Can be given.

【0018】

Furthermore, the present invention can repair long code identification data that is missing due to fading or the like by making the data composed of L repetitions of the orthogonal code by the M symbol wait for redundancy, and reduces search retries as a whole. As a result, the search time can be shortened.

【0019】

Further, the present invention needs to establish the initial synchronization in the mobile station, that is, the synchronization of the spread code sequence before establishing the carrier wave and the clock synchronization. High-speed search is possible by in-phase addition corresponding to the deviation.

【0020】

[Means for solving problems]

In the CDMA search method of the present invention, when establishing the synchronization of the spreading code before the carrier synchronization is established in the mobile station, a plurality of symbols (M) spread by the spreading code called a short code are prepared. Data in which an orthogonal code is formed using the polarities of the M symbols is used as a downlink signal, and when detecting the orthogonal code, a combination of polarities that the code can take over a plurality of symbols (M) constituting the code. It is characterized in that in-phase addition is performed by a correlator having.

【0021】

Further, according to the present invention, in a CDMA system in which an asynchronous system between base stations is realized by a code called a long code peculiar to a base station, the data composed of the orthogonal code includes information for identifying a base station, and the information is used. It is characterized by narrowing down long codes or specifying long codes.

【0022】

Further, the present invention is characterized in that one code in the orthogonal code is used as a header.

【0023】

Further, the present invention adds M symbol data received by the combination of polarities determined by the orthogonal code in the I system and the Q system, respectively, and specifies the orthogonal vector received with the sum of squares or the maximum value of the amplitude component. It is characterized in that the long code is narrowed down or identified based on the data obtained by repeating the L times.

【0024】

Furthermore, the present invention has an M-tap ring buffer that can be added with the polarity corresponding to the orthogonal code in each of the I system and the Q system, updates the absence of the ring buffer at the symbol rate, and identifies the addition result as an orthogonal vector. It is configured to be used.

【0025】

Further, the present invention is characterized in that a barker series is used to specify the frame position of the M symbol.

【0026】

Further, the present invention is characterized in that the above function is applied to a serial search, a parallel search, or a serial / parallel mixed search.

【0027】

Furthermore, the present invention to have a redundancy in the code formed by the L iterations of orthogonal vectors, the machine performs the repair locations that could not be detected constructed that it has added the capability as a feature.

【0028】

Further, in the present invention, when in-phase addition of M symbols is performed, a series in which weighting and polarity are performed assuming a carrier frequency deviation is prepared, and the maximum one is selected by observing the correlation, thereby achieving in-phase addition with frequency deviation. It is characterized by being able to respond.

【0029】

Further, the present invention is characterized in that, when preparing a series in which weighting and polarization are performed assuming a carrier frequency deviation in the case of in-phase addition of M symbols, approximation is performed only by the polarity.

【0030】

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, examples of the present invention will be described in detail with reference to the drawings.

【0031】

FIG. 1 is a first embodiment using the high-speed cell search method in the CDMA communication method according to the present invention.

【0032】

In the figure, the received signal after A / D conversion (omitted in the drawing) is multiplied by the diffusion signal generation circuit 1-1 with a phase shift function and the multiplier 2-1 to perform despreading. The result of this multiplication is stored in an accumulator composed of a buffer for one signal of the adder 4-1 and the ring buffer 3-1. As a result, the function of performing time integration in the unit signal longitudinal time period before diffusion is realized. Since the signal before diffusion does not change during the unit signal section, the above-mentioned time integration result of backdiffusion becomes the signal before diffusion.

【0033】

The timing chart of this situation is the first part of Fig. 3. In the figure, the time integration per unit time is accumulated with time, and the noise component is also integrated, but due to statistical independence, the increase in noise component is less than the increase in signal component, and the SNR is improved by that amount. To.

【0034】

Returning to Fig. 1, the operation of time integration over a unit signal as described above is performed in the same way for the next signal, but at that time the ring buffer rotates and the integration is performed in a new buffer. Results are accumulated. The operation is similarly performed for M of the ring buffer, and when the final integration operation is completed, symbol addition is performed for the contents of each buffer according to the polarity shown in FIG. Here, FIG. 2 shows an example of the relationship between the polarity of symbol addition and the orthogonal code, and shows the case of M = 4. When the same thing is done for each of the orthogonal codes shown in FIG. 2, the addition result is maximized when the pattern of M signals before diffusion and the combination pattern of polarities match. Figure 3 shows the state of time integration in this matched state. In the figure, the parts indicated by and arrows indicate the integration results stored in the buffer. It shows how the SNR is improved by adding M times. Here, FIG. 3 is a diagram showing the state of the signal by time integration and in-phase addition around the unit signal. By the way, as shown in FIG. 2, corresponding data is assigned to each Walsh-Hadamard code, and it is possible to discriminate the corresponding data by detecting the maximum value. By this operation, that is, in-phase addition, the data stored in each buffer sets the noise component of each buffer to X.₁ , X₂ , X_M Then, the power of the addition result is represented by the number 1 by its statistical independence.

【0035】

[Number 1]

<img file="JP2861985B2_D0001.tif" />Where the random variable X₁ , X₂ , X_M Have the same probability distribution, and the representative is X.

【0036】

On the other hand, the power of the addition result of the signal S is M.² S² Therefore, when converted to a level, the signal component is multiplied by M and the noise component is M by adding M times.^1/2 And the SN ratio is M^1/2 It can be seen that it is doubled and improved.

【0037】

In the case of the CDMA communication method, the noise component is larger than the signal component, and the expected SN ratio cannot be obtained only by performing time integration equivalent to the unit signal duration before diffusion. SN ratio by M^1/2 It can be doubled, and it is possible to realize the judgment for synchronous acquisition of the diffused signal.

【0038】

Next, the operation of synchronously capturing the diffusion code based on this determination result will be described. In Fig. 1, in order to shorten the search time, the serial search and the parallel search are mixed with a number of k parallel searches. In the figure, the kth block configured in parallel is shown, but k blocks with the same configuration are lined up to form one search circuit. However, the diffusion signal generation circuits 1-1 to 1-k with a phase shift function each have a constant phase shift, and the addition start position with respect to the unit signal of the ring buffer 3-1 to 3-k also shifts accordingly. doing. Therefore, the diffusion code can be synchronously captured by the maximum value of the correlation value corresponding to each correlator.

【0039】

FIG. 6 is an overall view when the CDMA cell search circuit of the present invention is applied to orthogonal detection. In the figure, 10-1 and 10-2 use the same cell search circuit as that used in the first embodiment of FIG. Selectors 10-1 and 10-2 are for selecting the additional output of the ring buffer, and are selected when each ring buffer is full. At this time, if the phase shift within the unit signal length + the phase shift amount of the unit symbol length is given to the timing of each ring buffer addition and the diffusion code for k ring buffers, it is useful for load distribution of the processing after the selector. .. The multipliers 12-1 and 12-2 and the adder 15 are for detecting the power by taking the sum of squares of I and Q respectively. The shift register 13 and the diffusion signal generation circuit 14 with a phase shift function are for supplying a diffusion code for reverse diffusion to the multipliers in the cell search circuits 10-1 and 10-2. In the case of this embodiment, the diffusion code is supplied. , I system, Q system use BPSK diffusion using the same diffusion code, so the diffusion code a from the shift register₁ ~ a_k Is shared. Also, a₁ ~ a_k Since the same code is only phase-shifted, the phase is changed and supplied by the shift register.

【0040】

After M synchronous additions in this way, the signal converted to power by the square addition of I and Q is subjected to power addition over the L frames by the L frame power adder 16, and is affected by ray-fading. Is mitigating.

【0041】

At this time, it goes without saying that instead of power conversion, an absolute value circuit is used instead, and the circuit scale is appropriately reduced.

【0042】

The orthogonal code is received in this way, and its frame configuration is shown in FIG.

【0043】

The present invention relates to a high-speed cell search in an asynchronous system between base stations, and each base station uses a different type of spreading code. Therefore, all kinds of diffusion codes are searched in the process of initial synchronization, but if it is done as it is, the search time will increase. Therefore, a high-speed search is performed using a long code mask symbol diffused only by a short code. The hatched part in the frame configuration shown in Fig. 4 is the long code mask symbol.

【0044】

In this embodiment, the orthogonal code composed of the M symbols is periodically sent to represent information for narrowing down (or specifying) one long code in the L frame. One of the L frames is used as a header for that information, so (M-1)^L-1 It is possible to identify individual long codes. In addition, the orthogonal code uses a part of the frame, and the correlation appears at the symbol spacing in the long code mask symbol. Therefore, it is possible to improve the accuracy of identifying the position of the M symbol in the frame by using a barker series with a sharp autocorrelation. Figure 4 shows the frame configuration at that time. When these M orthogonal code L frames are detected, the short code timing, long code timing, and long code identification can be known, so long code correlation detection can be performed based on this, and the long code mask symbol can be detected. Long code candidates can be narrowed down at an early stage by transmitting information on long code candidates.

【0045】

Here, since the cell search immediately after the power is turned on is the initial synchronization, it is necessary to detect the cell search with the phase shift of the carrier. Therefore, the polarity of the received symbol is unknown. However, since the present invention allocates orthogonal vectors such as the Hadamard matrix and the Weorsh matrix of the M symbol as the M symbol, it can be detected even with the phase shift of the carriers. The positions of the L M symbols can be assigned to each frame as shown in Fig. 4, or can be arranged collectively within one frame.

【0046】

FIG. 5 shows how the above-mentioned long code is identified. In FIG. 5, the first ν section indicates a section in which the short code is synchronized, and the first orthogonal code is detected in this section. Next, this is an example in which the remaining L-1 orthogonal codes are detected in the L sections, and the long code is finally identified from the obtained information.

【0047】

From this figure and Fig. 1, the time for inspecting the state of the entire 1-frame cycle when performing a cell search in which serial and parallel are mixed is shown in Equation 2.

【0048】

[Number 2]

<img file="JP2861985B2_D0002.tif" />Here, the short code period is P, the number of parallel processes of the correlator is C, the number of samples per chip is S, and the frame length is 10 ms.

【0049】

Therefore, the number of parallel processes C can be flexibly selected according to the limitation due to the circuit scale and the required search time.

【0050】

It is also possible to improve the fading resistance by adding redundancy to the long code information by the orthogonal code of the M symbol and repairing the part that could not be detected by fading during detection of L pieces.

【0051】

As an example, when ν = 8 (80ms) and L = 10, the search time is (8 + 10) × {10ms} = 180ms.

【0052】

FIG. 7 is a block diagram of all cell searches including long code identification. In the figure, the long code timing detection and the long code candidate detection 20 detect the type and position of the orthogonal code, that is, the timing of the long code, based on the combination of the polarities of the orthogonal code described above. The correlation unit 21 has a short code generation circuit inside, and outputs the correlation value obtained by the correlation device 29 to the ring buffer 23 for common mode addition by RAM. The ring buffer by RAM realizes the same function as in Fig. 1 in a form that reduces the hardware scale, and the orthogonal code control circuit inside the maximum value detection circuit 22 adds in-phase to the combination of polarities that can be taken by the orthogonal code. To do. The in-phase addition values of the I system and the Q system obtained as a result are converted into power by the sum of squares circuit 24, and the result is output to the maximum value detection circuit 22. The maximum value detection circuit 22 has a chip counter and a symbol counter, and stores the chip position having the maximum value, the symbol position, and the polarity of the orthogonal code at that time in the maximum value register. The final remaining value is output to long code identification 25. The long code identification 25 converts data from the orthogonal code based on the information sent from the long code timing detection and the long code candidate detection 20, and detects the long code narrowing down information. Based on the narrowed down information, the long code unique to the base station is identified. The narrowed down information is sequentially sent to the long code generation circuit 26, and the correlation value is detected by the long code correlator 28. As for the detection result, the maximum value determination circuit 27 determines that the most correlated code is a long code.

【0053】

Figure 8 shows the details of the ring buffer for common mode addition by RAM. The information back-spread by the matched filter or the correlator 31 of the serial / parallel mixture is output to the ring buffer 33 for common mode addition by RAM. The ring buffer for common mode addition by RAM can freely set the polarity to be added by external control, and outputs the common mode addition value according to the polarity for each of the I system and Q system. The maximum value of the sum of squares is detected by the maximum value detection circuit 32.

【0054】

Figure 9 shows the in-phase addition corresponding to the case where there is a frequency deviation between the carriers of the mobile station and the base station. When adding symbols, the lead frequency (f + δf), synchronization frequency (f), and lag frequency (f-δf) are set. It is corrected by the corresponding rotation phaser. It is also possible to set the correlation comparison target accurately by the ROM table of sin and cos. However, Fig. 9 shows an example in which a correlation comparison target is provided in which the frequency deviation is corrected by a simple circuit without using a ROM table. It is set to 2 and realized by addition and subtraction, respectively.

【0055】

Figure 10 shows an example when m = 4 and the polarity by the orthogonal code is ++++. From the figure, it can be realized by correlation comparison addition and subtraction of in-phase addition at carrier frequency deviation.

【0056】

[Effect of the invention]

As described above, if the high-speed cell search method for CDMA of the present invention is used, long codes can be identified or narrowed down by repeating M long code mask symbols L times, and at the same time, they are in phase over the M symbols. Since addition is performed, it is possible to narrow down long code candidates at an early stage, shorten the initial synchronization time, and provide a good CDMA system.

【0057】

Further, in the present invention, it is necessary to detect that rake synthesis cannot be used in the initial synchronization immediately after the power is turned on, that the power is dispersed by multipath and the signal component after despreading is low, and that the carrier is still accompanied by a phase shift. It is very effective in shortening the search time even in a harsh environment peculiar to mobile communication.

【0058】

Further, the present invention can be applied to orthogonal demodulation because it is possible to determine the orthogonal sign by taking the sum of squares of I and Q after adding in-phase to each of the I system and the Q system to increase the SNR. ..

【0059】

Further, in the present invention, the content is updated for each symbol rate with an M-tap ring buffer capable of in-phase addition with the polarity corresponding to the orthogonal code, and the orthogonal code is specified based on the addition result, so that the hardware scale is reduced and the consumption is low. It is possible to provide a search circuit suitable for power consumption.

【0060】

Further, since the present invention uses a sharp code of autocorrelation such as a barker series for specifying the frame position of the M symbol, it is possible to shorten the search time without erroneously recognizing the correlation appearing for each symbol. You can.

【0061】

Further, the present invention can be applied to serial search / parallel search or serial / parallel mixed search, and the degree of parallelism, that is, the search time can be selected according to the feasible circuit scale, which makes the design flexible. You can have it.

【0062】

Further, according to the present invention, the long code identification data lost due to fading or the like can be repaired by providing redundancy in the data composed by repeating the orthogonal code L times by the M symbol, so that the number of search retries can be reduced. As a whole, the search time can be shortened.

【0063】

Further, the present invention needs to establish the initial synchronization in the mobile station, that is, the synchronization of the spread code sequence before establishing the carrier wave and clock synchronization, but the frequency is used in a relatively small circuit even when there is a carrier frequency deviation. High-speed search is possible by in-phase addition corresponding to the deviation.

[Simple explanation of drawings]

[Figure 1]

It is a figure which shows the 1st Example which used the high-speed cell search system in the CDMA communication system by this invention.

[Figure 2]

It is a figure which shows the example of the relationship between the polarity of symbol addition and the orthogonal code.

[Fig. 3]

It is a figure which showed the state of a signal by time integration per unit signal and common mode addition.

[Fig. 4]

It is a figure which showed the frame composition structure.

[Fig. 5]

It is a figure which shows the timing chart when the long code was identified by repeating L times of the orthogonal code of the M symbol.

[Fig. 6]

It is a cell search circuit when this invention is applied to orthogonal detection, and is the figure which shows the 2nd Example.

[Fig. 7]

It is a block diagram of the whole cell search including the long code identification circuit.

[Fig. 8]

It is a figure which shows the search circuit with the common mode addition function using RAM.

[Fig. 9]

It is a figure which showed the symbol addition polarity correction rotation phaser of the in-phase addition corresponding to the carrier frequency deviation.

[Fig. 10]

It is a figure which shows the symbol addition polarity correction table of the carrier frequency deviation correspondence common mode addition.

[Explanation of symbols]

1-1 ~ 1-k Diffusion signal generation circuit with phase shift function 2-1 ~ 2-k multiplier 3-1 ~ 3-k ring buffer 4-1 ~ 4-k adder 5-1 ~ 5-k adder 10-1, 10-2 Cell search circuit 11-1, 11-2 selector 12-1, 12-2 squared circuit 13 shift register 14 Diffusion signal generation circuit with phase shift function 15 adder 16 L frame power adder 20 Long code timing detection and long code candidate detection 21 Correlation part 22 Maximum value detection circuit 23 RAM-based ring buffer for common mode addition 24 Square sum circuit 25 Long code identification 26 Long code generator circuit 27 Maximum value judgment circuit 28 Long code correlator 29 Correlator 31 Matched filter or serial / parallel mixed correlator 32 Maximum value detection circuit 33 RAM-based ring buffer for common mode addition