JP2010278829A - Software radio set - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a software radio set which performs communication corresponding to a time restriction determined by the standards even if a delay time is varied during execution. <P>SOLUTION: The software radio set includes, for example: a reception processing section 12 for executing predetermined baseband processing utilizing software; and a progress management section 11 for confirming, in a predetermined timing, the progress of the baseband processing being executed by the reception processing section 12 and determining processing contents to be executed in the subsequent baseband processing by the reception processing section 12 on the basis of a progress confirmation result. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a software defined radio.

  In recent years, software / processor-based processing systems have attracted attention in wireless communication baseband processing (see, for example, Patent Document 1 below). In such a processing system, since processing is realized using software, there are advantages such as improvement of design efficiency and easy correspondence to a plurality of wireless standards (communication standards).

  For example, if a single general-purpose processor does not have enough processing power in an implementation using a processor, a special processor is used, a multiprocessor configuration is used, and some processing is performed using dedicated hardware blocks. A desired radio is realized by using a method such as Hereinafter, the wireless device realized in this way is referred to as a “software wireless device”.

  Here, one of the things that must be taken into account when realizing the above-described software defined radio is a processing delay time. In general, when a delay time is defined in each corresponding wireless standard in wireless signal processing, it must be observed. For example, in a wireless LAN system, ACK must be returned within SIFS (Short Interframe space) after receiving a signal. For this reason, received signal processing must be completed with a delay time that allows an ACK to be returned within SIFS (= 16 μs). However, when performing signal processing with a software defined radio, special consideration must be given to this delay time. Necessary. This is because, unlike the implementation using dedicated hardware, the processing delay time may vary depending on the execution situation.

  When signal processing is created with dedicated hardware, processing is basically completed with a delay time designed by the system designer, and the delay time does not vary. However, in the case of a software defined radio, the delay time varies depending on various factors such as cache miss and resource contention, depending on the configuration. For example, when synchronization between processors is required in the implementation of a multiprocessor configuration, the delay time may vary depending on the time required for synchronization. This is because the time required for synchronization between processors is generally not determined statically. As another example, when a processor or hardware block is connected by a shared bus, a hardware block to be waited for appears when a bus conflict occurs. Since bus contention may or may not occur depending on the usage of the bus, it cannot basically be estimated statically. In addition, in the case of a system implementation in which a processor allocates processing to various hardware blocks and accepts an interrupt when processing is completed, there is a possibility that multiple interrupts may occur and low priority interrupt processing may be waited for . As described above, depending on the configuration of the software defined radio, there is a variation factor of the delay time that is not included in the radio realized by dedicated hardware.

  Therefore, in designing software radios, if the system is designed without considering the processing delay that appears at the time of execution as described above, there is a possibility that the time limit of the wireless standard cannot be observed due to the extended delay time at the time of execution. come. For example, in the case of a wireless LAN, ACK cannot be returned within SIFS due to processing delay, and if the receiving side does not return ACK, it is treated as reception failure on the transmitting side. For this reason, it is desired to realize a system that can observe the processing delay that appears at the time of execution and comply with the time constraints of the wireless standard. Conventional software defined radios mainly consider the processing delay that appears at the time of execution as a margin. The design method was adopted. For example, when the processing delay margin derived from a certain wireless standard is 16 μs, if it is expected that a processing delay of 2 μs will occur at the time of execution of processing for this timing budget of 16 μs, the processing is designed to finish within 14 μs. , A technique like this was adopted.

  However, when the design method as described above is adopted, if a processing delay exceeding the considered delay amount (margin) occurs, it cannot be dealt with and the time constraint specified by the system cannot be satisfied. . On the other hand, it is possible to avoid the processing delay exceeding the margin by designing the margin assuming the worst case where all delay factors overlap, but the timing budget becomes very small and the design becomes difficult.

JP 2004-40256 A

  An object of the present invention is to provide a software defined radio that performs communication corresponding to the time constraint defined by the standard even when the delay time varies during execution.

  According to one aspect of the present invention, a signal processing unit that executes predetermined baseband processing using software, and the progress of the baseband processing that is being executed by the signal processing unit are confirmed at a predetermined timing. There is provided a software defined radio including a progress management unit that determines the processing content to be executed by the signal processing unit in the subsequent baseband processing based on the confirmation result.

  According to the present invention, even when an unexpected delay occurs during the execution of a process, the generated delay can be canceled and the process can be terminated as scheduled. Therefore, even if the delay time fluctuates, it is possible to perform communication corresponding to the time constraint defined in the wireless standard, and the margin of processing delay secured at the time of design can be reduced as much as possible. Play.

FIG. 1 is a diagram illustrating a system configuration example of a software defined radio. FIG. 2 is a functional block diagram of the software defined radio. FIG. 3 is a diagram for explaining the operation of the software defined radio. FIG. 4 is a diagram for explaining the operation of the software defined radio. FIG. 5 is a diagram for explaining the operation of the software defined radio. FIG. 6 is a diagram for explaining the operation of the software defined radio. FIG. 7 is a diagram for explaining the operation of the software defined radio. FIG. 8 is a diagram for explaining the operation of the software defined radio. FIG. 9 is a diagram for explaining the operation of the software defined radio. FIG. 10 is a diagram for explaining the operation of the software defined radio. FIG. 11 is a diagram illustrating an internal configuration example of a reception processing unit.

  Hereinafter, a software defined radio according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. In addition, this invention is not limited by this embodiment.

(Embodiment)
In this embodiment, as an example, the reception operation of the software defined radio will be described. FIG. 1 is a diagram illustrating a system configuration example of a software defined radio according to the present embodiment.

  As shown in FIG. 1, the software defined radio of this embodiment includes processors 1 and 2, a hardware engine 3 that performs specific signal processing on an input signal, and FIR filter processing on the input signal. Filters 4 and 5 are provided, and each of these components has an internal memory (not shown) and is connected to the bus 6. Note that FIG. 1 shows an extracted portion necessary for realizing the characteristic operation of the software defined radio for the sake of simplicity. In the software defined radio according to the present embodiment, when a signal transmitted from the opposite communication device (communication partner) is received, baseband processing using software is performed to restore the transmission signal. Specifically, first, when signals are received by two systems of antennas (not shown), the analog signal processing unit (not shown) performs predetermined processing on each received analog signal, and the result is Filters 4 and 5 perform FIR filter processing on each of the two systems of digital signals generated based thereon. Next, the processors 1 and 2 and the hardware engine 3 perform predetermined digital signal processing (baseband processing) including demodulation processing and decoding processing on each signal obtained by performing the FIR filter processing. The data series transmitted is restored.

  FIG. 2 is a functional block diagram of a software defined radio having the architecture shown in FIG. As illustrated, the software defined radio according to the present embodiment includes a progress management unit 11, a reception processing unit 12, and a time measuring unit 13. In FIG. 2, as in FIG. 1, constituent elements necessary for realizing a characteristic operation are extracted and shown. Further, in this embodiment, in order to describe the reception operation, functional blocks that are not related to the signal reception operation are omitted.

  The progress management unit 11 manages the progress of processing executed by the reception processing unit 12. The reception processing unit 12 constitutes a signal processing unit (not shown), and when a signal received from the opposite communication device is input, predetermined digital signal processing (baseband processing) including demodulation processing and decoding processing Execute. That is, the reception processing unit 12 is realized by the processors 1 and 2 and the hardware engine 3 illustrated in FIG. The timer unit 13 manages time. The progress management unit 11 and the time measuring unit 13 are realized by the processor 1 and / or the processor 2.

  Subsequently, characteristic operations of the software defined radio having the above-described configuration will be described with reference to FIGS. 1 and 2 and FIGS. 3 to 10 as necessary. 3 to 10 are diagrams for explaining the operation of the software defined radio. The software defined radio according to the present embodiment differs from the general software defined radio in the following points. In other words, the software defined radio according to the present embodiment uses a clock or the like to monitor how much time has passed and how much processing has been performed while signal processing is being performed. And if the progress of the process is behind the speed assumed by the system (when it is detected that the delay has occurred due to the delay factor that appears at the time of execution), a part of the process by the software is changed to increase the processing speed. To make up for the delay. For example, when the reception processing unit 12 illustrated in FIG. 2 executes the reception processing, the progress management unit 11 calculates the scheduled end time when the series of reception processing in the reception processing unit 12 proceeds smoothly without delay. In addition, the progress of processing in the reception processing unit 12 (processing progress status) is appropriately checked, and reception is performed based on the result of this confirmation and time information obtained from the time measuring unit 13 (information on the time of confirmation). It is determined whether or not the process is finished by the scheduled end time. If it is determined that the process will not be completed by the scheduled end time, the progress management unit 11 instructs the reception processing unit 12 to change a part of the process to increase the processing speed. When receiving a change instruction from the progress management unit 11, the reception processing unit 12 changes a part of the process according to the content of the instruction and speeds up the reception process.

  This operation will be described as an example in the case of operating as a communication device on the receiving side of the wireless LAN system. In the wireless radio frame reception process, first, the reception processing unit 12 executes the header part reception process to obtain information on the data length (L) of the data part. When the information about the data length L is obtained, the progress management unit 11 calculates how long the reception process for the entire frame is completed based on the data length L and the processing capability of the reception processing unit 12. Let T be the calculated time when the reception process ends. Here, if it is assumed that the ACK cannot be returned within SIFS (specified time) when the end time of the reception process is later than the time T, the reception processing unit 12 needs to complete the entire reception process by the time T. There is.

  This situation will be described with reference to FIG. A straight line 30 in FIG. 3 represents the progress of processing when the reception processing unit 12 of the software defined radio proceeds smoothly without any fluctuation in delay time during execution. That is, FIG. 3 shows the progress status of the operation when there is no delay appearing at the time of execution, and represents that reception processing takes the data length L only at time T.

  When the calculation process of the time T (scheduled end time T of the reception process) based on the data length L is completed and then the reception process of the data portion is started, the progress management unit 11 obtains from the time measurement unit 13 in the software defined radio. The received time information is used to monitor how much data the reception processing unit 12 has processed. For example, if the data portion of a frame handled in a wireless LAN system is processed in units of blocks called symbols, each time a predetermined number of symbols (for example, one symbol) is processed, the current time is recorded. Monitor the progress of the process. Specifically, for example, the reception processing unit 2 notifies the progress management unit 11 every time processing for a predetermined number of symbols is completed, and the progress management unit 11 that has received this notification Information on the time at the time is acquired from the time measuring unit 13. Then, based on the acquired time information and the scheduled end time T, the progress management unit 11 determines whether or not the reception process in the reception processing unit 2 is completed by the scheduled end time T.

  If the reception process proceeds smoothly without delay in execution (as estimated in advance), the time obtained at the end of each symbol will be on the straight line shown in FIG. 4). The points shown in FIG. 4 represent the progress of processing in the reception processing unit 12 when the software defined radio is actually operated. For example, the result of plotting the time measured each time processing for one symbol is completed. It is. If a delay occurs for some reason, these points (plot results) are as shown in FIG. 5 and are shifted to the right from the straight line. That is, due to the delay that occurred during the execution of the reception process, it is delayed compared to the case where the process proceeds smoothly. If the processing is continued as it is, as shown by the dotted line in FIG. 5, the scheduled end time T is not in time even if the subsequent processing proceeds smoothly. Therefore, in such a case, the software defined radio accelerates the reception process by the following method and operates in time for the scheduled end time T.

  For example, the reception processing unit 12 of the software defined radio increases the processing speed by extending the correction amount update interval and reducing the processing load in the equalize processing. Equalize processing is processing that corrects the received signal affected by the communication channel by numerical calculation, but generally the wireless communication channel changes with time, so the correction amount is also updated according to the change of the wireless communication channel. It is necessary to continue. By omitting this correction amount update process (for example, by thinning out the update process executed periodically to widen the update interval), the calculation load at the time of reception can be reduced and the speed can be increased. Thus, by extending the update interval of the correction amount of the Equalize process, the speed of the reception process is increased, and the state where the reception process shown in FIG. 5 is delayed is shown in FIG. It can be improved until the point indicating is on the thick line 60. FIG. 6 shows an example in which the process is switched to the point A and the process is switched to speed up the reception process so that the scheduled end time T is met.

  Although an example in which the processing speed is increased in the reception operation of the wireless LAN system and the processing is ended by the scheduled end time T has been described here, various variations are conceivable. In other words, the overall processing speed can be increased by a similar method (reducing the load on a predetermined process) in the transmission operation and the signal processing operation (transmission / reception operation) in another communication system.

  Further, the characteristic operation of the software defined radio according to the present embodiment is composed of the following three elements (step S1) to (step S3).

(Step S <b> 1): The progress management unit 11 acquires “the set of time and the amount of work completed by that time”.
(Step S2): The progress management unit 11 determines the progress of the process (whether it is delayed or advanced from the scheduled end time) based on the information acquired in step S1.
(Step S3): If it is determined in step S2 that it is “delayed”, the processing is switched to high-speed processing. That is, the progress management unit 11 instructs the reception processing unit 12 to accelerate the processing speed, and the reception processing unit 12 accelerates the processing speed according to the content of the instruction.

  Each of these steps will be described in detail. First, when acquiring “a set of time and work amount completed by the time” in step S1, the progress management unit 11 inquires of the time measuring unit 13 about the time and acquires time information. The “time information” is a time from an appropriate time point such as 0 at the head position of the frame. The “amount of work completed by the time” may be anything as long as it is information that quantitatively indicates the amount of work. For example, the amount of data that has been processed (the number of bytes, the number of symbols, etc.) may be used. In addition, when iterative decoding is performed, information indicating how many iterations have been completed may be set as “the amount of work completed by the time”. Note that even if the amount of data at the time when some processing is completed, there is a choice as to which “processing” the data amount at the time when the processing is completed. For example, when the number of processed symbols is used, if the reception process is executed in a pipeline manner in the order of (1) FFT (Fast Fourier Transform) process, (2) Equalize process, and (3) Demapping process, It is necessary to select an appropriate position, such as whether the number of data symbols for which (1) has been completed, whether the number of data symbols for which (2) has been completed, or the number of data symbols for which (3) has been completed. The progress management unit 11 acquires such work amount information from the reception processing unit 12.

  The following is an example in which the process of step S1 is implemented. If you want to know the progress of processing executed by hardware, prepare a counter that counts the amount of data processing, and when that counter reaches a predetermined value, that is, processing for a certain amount of data (for example, one symbol data) The progress management unit 11 obtains time information from the time measuring unit 13 at the time when is completed, thereby obtaining a set of time and data processing amount. If it is desired to know the progress of processing executed by software, the progress management unit 11 acquires time information from the time measuring unit 13 at the end of processing (for example, when processing by a specific function is completed). Good. In addition, if the reception process is configured with hardware and a processor that controls the hardware, the processor can know that the predetermined processing has been completed by receiving an interrupt from the hardware. Whenever the interruption is received, the progress management unit 11 acquires time information from the time measuring unit 13, whereby time information at the time when the processing up to that point is completed can be obtained.

  In step S2, the progress of the process is determined based on the information (a set of time and finished work amount) acquired in step S1, and this determines the point A shown in FIG. 6 (reducing the process). To determine the timing of speeding up). The implementation method of step S2 is roughly divided into the following two.

  In the first method (method 1), a deadline of each process is obtained in advance by calculating backward from the scheduled time when a series of reception processes must be completed, and the actual progress of the process is determined from the deadline. If it is late, switch the process.

  In another method (method 2), the expected progress of the process is determined in advance, and if the actual process determines a delay of a certain amount or more (if a delay exceeding the allowable range occurs), the process is performed. Switch.

A specific example of Method 1 will be described. As described in the example of the wireless LAN system, in many communication systems, the standard determines that ACK is returned within a certain time from the time when data flowing through the communication path ends ( _Tend ). ing. The time T _{end when the} data flowing through the communication path ends can be calculated by decoding data length information included in the received data (received frame). The data length information is not limited to the case of the wireless LAN system, and is generally placed at the beginning of the frame. Therefore, T _end can be calculated at an early stage after the reception of the frame is started. Therefore, the software defined radio must return the ACK within a predetermined time determined by the standard based on the time T _end and considering the processing time for transmitting the ACK. It is possible to determine whether the reception process has to be completed, that is, the reception process end scheduled time T (for example, T <T _end + SIFS). When calculating these T _end and T, in the software defined radio, when the decoding process for the data length information is completed, the reception processing unit 12 notifies the progress management unit 11 of the data length information, and the progress management. The unit 11 calculates T _end and T based on the notified information.

  After determining the scheduled end time T in this way, the software defined radio changes to a process with a light load when a delay is detected, as in the above-described example of the wireless LAN system (examples shown in FIGS. 3 to 6). As a result, a timing (corresponding to the point A shown in FIG. 6) that is just before the scheduled end time T is obtained. That is, the process is switched when the progress of the process is about to go to the right side of the thick line 60 in FIG. The thick line 60 represents the reception processing capability (processing capability of the reception processing unit 12) of the software defined radio when the processing speed is increased by switching to a process with a light load. Even if it does, it cannot be in time for the scheduled end time T. Therefore, the point A is a point that will be in time for the time T if the process is switched.

  How to draw the thick line 60 depends on the situation. For example, the thick line 60 can be set as a processing capability when the processing proceeds smoothly (without a delay at the time of execution) after speeding up, or a delay that occurs at the time of execution to some extent after the speeding up is considered. It can also be an average processing capacity. In addition, if the deadline (corresponding to the above thick line 60) is drawn with a marginal line, it will not be in time if it is delayed even a little behind that line. Further, when obtaining the point A for switching the process, the overhead associated with the process switching should be taken into consideration. For example, when switching is performed by executing software, the overhead associated with the switching of processing is a delay due to branching or a delay due to switching of the contents of the instruction cache or instruction memory. Also, if there is a restriction that “the place where the process can be switched is only the symbol breakpoint”, it is not possible to immediately switch because the process is switched, and it is necessary to consider this.

  In FIG. 6, a straight line 61 indicating the normal processing capability of the software defined radio is also drawn. However, in the method of determining based on the deadline (thick line 60) as in Method 1, it corresponds to the straight line 61. No information is needed. Further, in FIG. 6, a plurality of “time and data processed amount” points (monitor results indicating the progress of reception processing) are drawn periodically, and it is determined whether each point is delayed. However, this point may be irregular (the period for executing the processing progress confirmation may be irregular).

  Next, a specific example of the method 2 will be described with reference to FIG. In FIG. 7, a straight line 72 represents a normal processing capability that does not include a delay at the time of processing execution. The straight line 70 is a straight line drawn from the straight line 72 in consideration of a certain amount of delay, that is, the maximum delay amount allowed in the system, and is parallel to the straight line 72. Each point in the figure indicates the progress of the reception process (monitor result). The maximum delay amount is determined so as to satisfy the time constraint in consideration of the time constraint derived from the communication standard corresponding to the processing to be executed. For example, it is determined when designing the software defined radio. In this method 2, when the point indicating the monitoring result is likely to go to the right side of the straight line 70, that is, when the point is delayed by a certain amount or more from the expected progress of the processing (straight line 72), the processing is accelerated. In the example of FIG. 7, the progress management unit 11 detects that the point B reaches the right side of the straight line 70 and is delayed by a certain amount or more from the schedule (straight line 72). Then, the progress management unit 11 instructs the reception processing unit 12 to switch to the processing with a light load, and the reception processing unit 12 changes the processing speed from the point B to the processing speed indicated by the straight line 71. The delay is recovered by accelerating.

  In the operation illustrated in FIG. 7, after the point B, the progress management unit 11 detects that the delay has been eliminated at the point C, and as a result, the reception processing unit 12 finishes the high-speed processing, and the straight line 72. Processing corresponding to is started. In addition, although it returned to the original process (process of the processing speed shown by the straight line 72) at the point C, how to select the position of the point C is arbitrary. Also, it is not essential to return to the original process, and if another process can be selected, it may be started.

  Unlike method 1 described above, this method 2 can be applied without obtaining the scheduled end time T, and thus can be used in a broadcast receiving system or the like. In the case of application to a broadcast receiving system, a straight line corresponding to the straight line 70 shown in FIG. 7 may be obtained based on a delay time required from subsequent processing of broadcast wireless processing (moving image decoding processing or the like).

  In the above methods 1 and 2, the progress is determined during the series of reception processing and the high-speed processing is switched. However, whether to switch to the high-speed processing is determined based on the past progress situation (operation result). It is also possible. In other words, when executing a certain process, if a delay occurred when the same process was executed last time, it is estimated that the same delay is likely to occur again, and the processing load of the process to be executed is reduced. Reduce in advance (execute processing faster than normal). For example, in the case of a wireless LAN system, the processing is switched such that “the processing of the previous frame took longer than planned, so this time high-speed processing is used”. Applying acceleration according to the previous progress, such as applying the fastest processing as possible this time if the previous progress is greatly delayed, and slightly increasing the speed this time if it is only a little late before. Is a way of changing. In this method, once switching to high-speed processing, there is a possibility that it will end earlier than expected. Therefore, when it is likely to end earlier than expected, the process is appropriately switched by returning to the original process.

  The above methods 1 and 2 and the method for determining whether to switch to high-speed processing based on past progress can be used in combination of two or more. For example, the control shown as the method 2 is performed halfway, and finally the control shown as the method 1 is applied in order to meet the scheduled end time T.

  As described above, when executing step S3 and speeding up the processing executed by the reception processing unit 12, if the reception processing is performed in the wireless LAN system, the processing load can be reduced by omitting part of the equalization processing. Various methods can be applied depending on the wireless standard and the wireless device mounting method. Mainly, (Method A) improves the operating frequency. (Method B) Increase computational resources. (Method C) A part of the processing is omitted. (Method D) The algorithm is changed. (Method E) Reduce the amount of processing data to be handled. Etc.

  (Method A) and (Method B) are methods for increasing the processing speed without reducing noise tolerance by increasing the calculation capability. The method of improving the operating frequency of (Method A) is realized by returning the voltage or frequency that has been lowered by DVFS (Dynamic Voltage Frequency Scaling) or the like for low power operation to the original operation. Further, it is possible to enable operation at a higher frequency by utilizing the substrate bias effect. By these methods, the frequency can be increased within a range that meets the critical path of the circuit. The realization of (Method B) is based on the premise that there are resources other than the currently used calculation resources. For example, in the case where some processor cores are powered down for low power operation, the processor cores are turned on and used to increase the calculation capability. In addition, even if a calculation resource is being used in another process, if the process being executed is not critical, a method of increasing the calculation capability by acquiring and using the resource can be considered. For example, if the current processing is delayed due to a conflict of bus requests, take a method such as lowering the priority of bus requests other than the processing you want to prioritize or not generating the request itself. It will be realized.

  (Method C), (Method D), and (Method E) are methods that speed up processing at the expense of noise tolerance. (Method C) is such that calculation for rounding error countermeasures is changed so as not to be performed by software, the number of repetitions in error correction processing for iterative decoding is reduced, error correction is not performed for systematic codes, etc. Realized by making changes. Further, (Method D) is realized by reducing the processing and increasing the speed by changing the Equalize method (Sphere Decoder, MMSE, Zero Forcinng). (Method E), for example, reduces the memory access and bus load by reducing the number of quantization bits, and when performing MIMO (Multiple Input Multiple Output) decoding in a multiprocessor configuration, passes information necessary for decoding. For example, by eliminating communication between processors.

  It is also possible to increase the speed by combining two or more of (Method A) to (Method E). When speeding up the processing, if there are a plurality of speed-up options (methods), it is necessary to select which processing to change. In general, speeding up leads to an increase in power consumption or deterioration in noise resistance, and there is a trade-off relationship between these and the realization of speeding up. Therefore, it is desirable that the software defined radio can adaptively determine which method is used for speeding up depending on the situation. For example, in a noisy environment, if you select a process that is too light and degrade the noise tolerance, you may not be able to receive (CRC error, etc.). The process to be changed should be selected in consideration of deterioration. In order to know the state of the communication channel, the general RSSI value, pilot signal power, EVM (Error Vector Magnitude), error correction decoder status (for Viterbi decoder, integrated path metric, turbo code, etc.) For decoders that perform iterative decoding, refer to the number of iterations until stopping), the amount of correction value during Equalize processing, etc.

  An example of switching processing according to the state of the communication path will be described with reference to FIGS. It is assumed that the reception processing unit 12 of the software defined radio usually has a reception processing capability indicated by a straight line 80 in FIGS. The reception processing unit 12 has two load reduction options: a processing speed operation indicated by a straight line 81 and a processing speed operation indicated by a straight line 82. The operation corresponding to the straight line 81 is an operation corresponding to the straight line 80. It is assumed that the load is lighter than that (normal operation at the processing speed indicated by the straight line 80) but the noise resistance is weak, and the operation corresponding to the straight line 82 is lighter than the operation corresponding to the straight line 81 and further has a low noise resistance. FIG. 8 shows an example of an operation result when the software defined radio performs a reception process in a state where the noise on the communication path is sufficiently small. Specifically, an example of the progress result of the process when an increase in delay is detected during the process and the process is switched to a process with a light load is shown. In the example shown in FIG. 8, the straight line 82 has the weakest noise tolerance, but since the noise in the communication path is sufficiently small, the processing satisfying the time constraint can be achieved while maintaining the receivable quality as a whole. On the other hand, FIG. 9 shows an example of an operation result when the software defined radio performs reception processing in a state where noise in the communication path is larger than that in FIG. Specifically, an example of the progress result of the process when an increase in delay is detected during the process and the process is switched to a process with a light load is shown.

  When performing the operations shown in FIGS. 8 and 9, in the software defined radio, the progress management unit 11 estimates the noise amount in the communication path with reference to the RSSI value that can be acquired from the reception processing unit 12, and which It is determined whether the quality that can be received as a whole can be maintained by changing the process to a process with a light load. For example, in the case shown in FIG. 9, if the progress management unit 11 switches to a process with a light load corresponding to the straight line 82 in the current communication path state, the quality deteriorates too much and a CRC error occurs. It is determined that the reception quality is to be maintained by performing at least processing corresponding to the straight line 81. Then, a delay occurs during the reception process, and the process is executed by the reception processing unit 12 when the process progress confirmation result (monitor result) is likely to go to the right side of the straight line 81. Switch to meet time constraints. By controlling in this way, it is possible to cope with a delay that occurs during processing execution while maintaining a certain level of reception quality.

  Which process is to be changed to a lighter one can be determined by paying attention to the characteristics and operating environment of each process. For example, in a software defined radio that normally operates at the processing speed indicated by the straight line 100 in FIG. 10, Equalize processing (processing # 1) and Viterbi decoding processing (processing # 2) are candidates for load reduction to increase the processing speed. ), If the process # 1 is changed to a process with a light load, the processing speed is switched from the straight line 100 to the straight line 101. On the other hand, if the process # 2 is changed, the processing speed is switched from the straight line 100 to the straight line 102. And In such a case, as can be seen from FIG. 10, the speed is increased only to the same extent regardless of which of the process # 1 and the process # 2. Therefore, although it may be switched to processing # 1 or # 2 in terms of acceleration, the progress management unit 11 determines whether to switch to processing # 1 or # 2 in consideration of the characteristics of the processing. For example, if the correction amount at the time of Equalize is large, it can be predicted that the quality degradation will be large if the algorithm is changed so that the Equalize process is not performed carefully by changing the process # 1, so the Viterbi decoding process of the process # 2 Speed up by reducing the load of On the other hand, when there is almost no correction amount at the time of equalization and the integrated value of the Viterbi decoding path metric is large, the processing # 2 is not changed, and the processing # 1 is changed to change the equalization processing.

  Furthermore, a method of selecting a process in consideration of the importance of data as to which process is to be changed when switching to a process with a light load can be considered. For example, since the reception process for the part containing the communication control information is important, a method of controlling the process so as to omit the process as much as possible can be considered.

  Finally, a specific example of changing the reception process to increase the processing speed will be described. FIG. 11 is a diagram illustrating an internal configuration example of the reception processing unit 12. As illustrated, the reception processing unit 12 includes FIR filters 21-1 and 21-2, AFC units 22-1 and 22-2, FFT units 23-1 and 23-2, a MIMO decoding unit 24, An arrangement unit 25 and a Viterbi decoding unit 26 are provided.

  The FIR filter units 21-1 and 21-2 perform predetermined processing on each analog signal received by two antennas (not shown), and perform FIR on each of the obtained two digital signals. Perform filtering. The AFC units 22-1 and 22-2 perform AFC (Automatic Frequency Control) processing on the signals input from the preceding FIR filters 21-1 and 21-2 to compensate for the frequency offset. The FFT units 23-1 and 23-2 perform FFT processing on the signals input from the preceding AFC units 22-1 and 22-2. The MIMO decoding unit 24 performs a MIMO decoding process including an Equalize process on the signals input from the FFT units 23-1 and 23-2. The MIMO decoding processing result output from the MIMO decoding unit 24 is a soft decision value. The arrangement unit 25 performs data exchange and alignment processing on the signal input from the MIMO decoding unit 24, and returns each data in the correct arrangement order. The Viterbi decoding unit 26 performs Viterbi decoding processing on the signal input from the arrangement unit 25.

  In the reception processing unit 12 having such a configuration, it is possible to reduce the processing load by reducing the number of quantization bits of the soft decision value in the MIMO decoding process, and to realize high-speed reception signal processing.

  The reason why the processing load can be reduced by reducing the number of quantization bits of the soft decision value will be described. Of the processing executed after the FIR filters 21-1 and 21-2 of the reception processing unit 12 described above, the processing in the Viterbi decoding unit 26 is performed by dedicated hardware (in the hardware engine 3 shown in FIG. 1). The other processing is executed by software (corresponding to the processors 1 and 2 shown in FIG. 1). When the processing is divided in this way, the amount of writing to the internal memory in the processors 1 and 2 can be reduced by reducing the number of quantization bits of the soft decision value. Therefore, the processing time in each processor is shortened and the processing speed is increased. In addition, since the amount of data transferred from the processors 1 and 2 to the hardware engine 3 is reduced, the bus occupation time is reduced, and the increase in processing time due to bus access contention is reduced. Furthermore, since the internal calculation accuracy of the Viterbi decoding process performed by the hardware engine 3 can be reduced, the operating frequency can be increased, and the time required for the error correction process can be shortened.

  However, there are points to be noted when reducing the processing load by reducing the number of quantization bits of the soft decision value. First, it is necessary to cope with the change in the number of bits of the soft decision value during decoding. For example, when the soft decision value is changed from 6 bits to 4 bits during decoding, the integrated value of the path metric is shifted by 2 bits, and it is necessary to correct it. Specifically, the path metric value before switching to 4 bits is set to 1/4, or after switching to 4 bits, the branch metric is multiplied by 4 and used. In addition, when a function for increasing the operating frequency by reducing the number of quantization bits is implemented, it is necessary to design the circuit in consideration of correct operation even after decreasing the number of quantization bits and increasing the operating frequency.

  As described above, the software defined radio according to the present embodiment monitors the progress of processing, and detects that the processing progress (speed) is delayed from the expected speed (delayed from the schedule due to a delay factor). In the event that it is detected that the process has been performed), a part of the process by software is changed to a process with a lighter load to speed up the process and to operate to recover the delay. Thus, even if an unexpected delay occurs during the execution of the process, the generated delay can be canceled and the process can be terminated as scheduled. As a result, even if the delay time fluctuates, it becomes possible to perform communication corresponding to the time constraint defined by the wireless standard, and it is possible to minimize the margin for the processing delay that is secured at the time of system design.

  1, 2 processor, 3 hardware engine, 4, 5 filter, 11 progress management unit, 12 reception processing unit, 13 timing unit, 21-1, 21-2 FIR filter, 22-1, 22-2 AFC unit, 23 -23-23 FFT unit, 24 MIMO decoding unit, 25 arrangement unit, 26 Viterbi decoding unit.

Claims (6)

A signal processing unit that executes predetermined baseband processing using software;
A progress management unit that confirms the progress of baseband processing executed by the signal processing unit at a predetermined timing and determines the processing content that the signal processing unit performs in the subsequent baseband processing based on the progress confirmation result When,
A software defined radio comprising: The progress management unit
Check whether the baseband processing is proceeding smoothly based on the communication standard corresponding to the baseband processing performed by the signal processing unit, and if not proceeding smoothly, 2. The software defined radio according to claim 1, wherein the software radio is instructed to switch the process being executed to a process with a light processing load. The progress management unit
The scheduled end time of the baseband processing being executed by the signal processing unit is calculated based on the communication standard, and every time the signal processing unit finishes a predetermined amount of work, from the start of the baseband processing. 3. The software defined radio according to claim 2, further comprising: determining whether the baseband processing is proceeding smoothly based on a confirmation result of the scheduled end time and the elapsed time. Machine. The progress management unit
Each time the signal processing unit finishes a predetermined amount of work, the elapsed time from the start of the baseband processing executed by the signal processing unit is confirmed, and determined based on the confirmation result of the elapsed time and the communication standard. 3. The software defined radio according to claim 2, wherein whether or not the baseband processing is proceeding smoothly is determined based on the maximum delay amount allowed in the system. The progress management unit
If the signal processing unit estimates whether or not the baseband processing to be executed in the future proceeds smoothly based on the past progress confirmation result and estimates that the baseband processing does not proceed smoothly, the signal processing unit 2. The software defined radio according to claim 1, wherein when the progress confirmation result is acquired, the software defined radio is instructed to execute a process having a lighter processing load than the process executed by the signal processing unit. The progress management unit further includes:
The software defined radio according to claim 5, wherein an instruction is given to switch processing executed by the signal processing unit in accordance with a state of a communication path.

Images