CN104486241B - Down channel data processing method in base band resource pool - Google Patents

Abstract

The present application discloses a downlink channel data processing method in a baseband resource pool, comprising: setting an exit buffer at the exit of the baseband processing in advance; selecting and sending the data processed by the baseband into the exit buffer; time interval, fetching data from the egress buffer for front-end processing, sequentially numbering and recording the data read from the egress buffer for front-end processing. By applying the present application, the real-time requirement of downlink channel data transmission can be guaranteed when downlink channel overtime congestion occurs in the baseband resource pool architecture.

Description

Downlink channel data processing method in baseband resource pool

technical field

The present application relates to wireless communication technology, in particular to a downlink channel data processing method in a baseband resource pool.

Background technique

In the traditional access network architecture, baseband signal processing is usually implemented on a dedicated processor using a fixed algorithm solution under the premise of meeting real-time requirements. The resources consumed are fixed, and the real-time performance of the solution is also determined. However, the base station resource pool platform architecture usually uses a general-purpose processor, and the operating system schedules multiple processing tasks. Compared with traditional platforms, the real-time performance of physical layer processing is difficult to guarantee, and the uncertainty will increase. Therefore, it is inevitable to encounter To the timeout congestion problem: The scheduling of other tasks by the processor causes the computing tasks of the baseband processing at the physical layer to be stagnated for too long, exceeding the time limit allowed by real-time performance, causing subsequent business data processing to be congested. This problem hardly exists in the traditional baseband processing process, so there is no solution to this problem.

Specifically, for traditional communication equipment, one manufacturer often provides a complete set of solutions, and there is a high dependence on system maintenance or upgrade. With the shortage of energy resources in recent years, global mobile communication network operators are facing increasingly serious cost pressures. Most mainstream operators usually have multiple networks with different communication standards. In order to ensure the quality of service of the network, a large number of base stations need to be deployed to solve the problem of network coverage. However, the relative scarcity of site and computer room resources creates a contradiction that is difficult to coordinate with the deployment of a large number of base stations. However, due to the fierce competition in the mobile communication market, the average revenue per user grows slowly or even declines, and the "profitability" of operators does not increase accordingly, which will lead to the compression of investment in network construction and equipment procurement. Considering the industry's sustainable profitability and long-term development, the mobile communication industry proposes to solve this problem by changing the access network architecture.

The architecture of the new base station system is shown in Figure 1. All baseband processing units (Baseband Unit, BBU) and remote radio frequency units (Radio Remote Unit, RRU) are connected through a high-bandwidth, low-latency optical transmission network. The baseband processing units are concentrated at one physical site to form a baseband pool. Multiple baseband processing units in the baseband pool are connected through high-bandwidth, low-latency, flexible topology, and low-cost cross-connections. The baseband resource pool needs to apply base station virtualization technology. In the baseband pool, multiple base stations share computing resources, and the allocation of computing resources is uniformly and dynamically scheduled by the system according to the traffic volume. The wireless signal processing algorithm constitutes the core processing of the physical layer of the wireless communication system, which is computationally intensive and faces strict real-time requirements. In order to ensure the real-time performance of centralized processing of base stations, reduce system energy consumption, and enable virtualization technology to maximize hardware system performance to support high-speed communication system baseband data processing, it is necessary to divide and package the computing tasks of baseband signal processing , to centrally process the encapsulated computing tasks. The encapsulation scheme can be adjusted according to the resource usage of the resource pool of the base station, so as to adapt to the dynamic task allocation on the multi-core general processor and meet the real-time requirements of the system.

Different from traditional base stations, the base station resource pool gathers a large number of computing resources to accommodate multi-cell multi-user data processing simultaneously. The real-time requirements of the physical layer signal processing of the wireless communication system are high. Taking the LTE system as an example, the downlink channel sends data to the user every 1 ms. The data of one subframe is sent, and the uplink channel receives the data of one subframe every 1ms. Since the resource pool platform architecture usually uses a general-purpose processor, and the operating system schedules multiple processing tasks at the same time, the physical layer processing in the resource pool will inevitably encounter the following problems: processor scheduling leads to the calculation of the physical layer baseband processing The stagnation time of the task is too long, exceeding the specified time limit, which causes the subsequent business data to be congested, as shown in Figure 2. This problem is referred to as "timeout congestion" in this application.

In Figure 2, the processor schedules other processing and computing tasks during the processing of the computing tasks of channel A, which causes the computing tasks of channel A to stagnate. If the channel is an uplink channel, due to the strong real-time processing of the receiving front end, The data cannot be processed in time and congestion occurs. If the channel is a downlink channel, due to the high real-time requirements of the sending front-end processing, the data may not be delivered to the front-end in time.

This problem hardly occurs on the dedicated processor used in the traditional baseband processing unit (BBU), but on the general-purpose platform, the operating system intervenes to schedule and process real-time processing tasks, and deal with the problem that real-time performance is difficult to guarantee due to overtime congestion. will be highlighted. However, many current congestion-handling technologies are often aimed at the problem of network packet congestion, and the corresponding solutions cannot be directly applied to this application to solve the problem of timeout congestion.

Contents of the invention

The present application provides a data processing method in the baseband resource pool, which can ensure the real-time requirement of data transmission when timeout congestion occurs in the baseband resource pool architecture.

In order to solve the above problems, the application adopts the following technical solutions:

A data processing method in a baseband resource pool, comprising:

a. Set the exit buffer at the baseband processing exit in advance;

b. The data after baseband processing is selected and sent to the egress buffer; according to the preset time interval, the data is taken out from the egress buffer for front-end processing, and the data read from the egress buffer is processed by the front-end The processed data is sequentially numbered and recorded;

Wherein, when there is no data in the export buffer, read out the empty data and continue to number and record the empty data;

The method of sending the data after baseband processing into the export buffer includes: if the processing output sequence number of the data is less than or equal to the maximum number of the currently recorded data for front-end processing, the data is discarded and not sent to the export buffer. the egress buffer; otherwise, send the data into the egress buffer;

The processed output sequence number of the baseband processed data is the sequence number when the data block is output after the baseband processing.

Preferably, the method between steps a and b further includes: assigning an equivalent processing output sequence number to the data before baseband processing, and the sequence number is the corresponding data block number when the input data of baseband processing is output after baseband processing sequence number;

When the timeout congestion occurs on the downlink channel and ends, estimate the timeout congestion degree O _M +δ-K _e , and when the timeout congestion degree is greater than or equal to 0, discard the data with equivalent processing output sequence numbers K _e to O _M +δ, No more baseband processing;

Wherein, _OM is the maximum serial number of the currently recorded data that undergoes front-end processing, K _e is the equivalent processing output sequence number of the data that is about to undergo baseband processing, and δ is a preset natural number.

It can be seen from the above technical scheme that a data processing method is provided in this application, an exit buffer is set at the exit of the baseband processing, and the downlink data read from the exit buffer or empty data are sent at fixed intervals. The processed downlink data that cannot be output on time is discarded. At the same time, the expired data that has not been processed in time by the baseband is also directly discarded, so as to ensure the real-time requirements of downlink data transmission and save baseband processing resources.

Description of drawings

FIG. 1 is a schematic diagram of a new base station system architecture;

Fig. 2 is a schematic diagram of overtime congestion;

FIG. 3 is a schematic flow chart of a data processing method for a downlink channel;

Fig. 4 is a schematic diagram of downlink channel data transmission and egress buffer positions;

Fig. 5a and Fig. 5b are respectively two example diagrams of query data output serial number table;

Fig. 6 is an example diagram of sending data and output sequence number table recording state when there is data in the export buffer;

Figure 7 is an example diagram of sending data and output sequence number table recording state when there is no data in the buffer;

FIG. 8 is an example diagram for judging expired data that has not been processed by the baseband.

detailed description

In order to make the purpose, technical means and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings.

The purpose of this application is to design a data processing method in a base station resource pool to deal with overtime congestion at the physical layer and ensure the real-time performance of the baseband signal processing process at the physical layer. The physical layer signal processing of the base station resource pool consists of multiple channels, including uplink and downlink channels. The uplink channel receives the signal of the user terminal from the air interface and then transmits it to the baseband for processing. This uplink transmission process has strict real-time requirements at the receiving end; while the downlink channel performs certain processing on the base station signal after baseband processing. It is sent to the user terminal through the air interface. This transmission process has strict real-time requirements at the sending end. If timeout congestion occurs during downlink channel processing, the data sent by the base station as the sending end may not be sent in time. If timeout congestion occurs during uplink channel processing, data received by the base station as the receiving end may not be processed in time. Both situations will affect The real-time performance of the physical layer processing of the baseband resource pool.

In view of the above-mentioned situation that the real-time performance may be affected after timeout congestion occurs on the downlink channel, this application provides a corresponding data processing method to avoid the real-time performance impact on the physical layer processing.

Specifically, corresponding to the case where the base station serves as the downlink channel sender, the present application provides the data processing method shown in FIG. 3 . Wherein, due to the high real-time requirements of the transmission of the downlink channel, this processing method is used to perform the processing of timeout congestion control at the exit of the baseband processing. As shown in Figure 3, the process includes:

Step 301, pre-setting an egress buffer at the baseband processing egress.

For the downlink channel, since the data after baseband processing and front-end processing needs to be sent to the user terminal through the air interface, as shown in Figure 4, real-time guarantee operations need to be performed at the baseband processing exit. Based on this, in this method, an egress buffer is used at the egress of the baseband processing, so that the data processed by the baseband is buffered and processed by the egress buffer and then transmitted to the front end. The egress buffer can be divided into multiple buffer blocks, and each buffer block stores a data packet output after baseband processing.

Step 302, before the baseband processing, assign an equivalent processing output sequence number to the data entering the baseband processing.

In the physical layer channel, there are various data processing units, such as transport blocks, code blocks, etc., and the data processing units entering the baseband processing are different from those at the output. For example, in the LTE system, the input data of the downlink channel is based on the transport block. Unit, the output after baseband processing is in frame, but there is a fixed conversion relationship between them. In order to facilitate the estimation of the overtime congestion degree in the subsequent steps, it is necessary to assign an equivalent processing output sequence number to the data entering the baseband processing. The sequence number is a number of natural counts, and the sequence number counted in units of output data units when the input data is output is obtained according to the relationship between the amount of input and output data. For example, m input data units correspond to one output data unit, and the same natural number number is assigned to every m input data units. Numbering adopts circular natural counting, n=1,2,...,N. The equivalent processing output sequence number assigned in this step will be used in step 305 to calculate the timeout congestion degree and related processing.

Step 303, assigning a processing output sequence number to the baseband processed data, and selectively writing the processed data into the egress buffer according to the sequence number.

In this step, the write operation of the export buffer is performed. In order to facilitate the real-time guarantee operation of the data, it is necessary to assign a processing output sequence number to the data block processed and output by the baseband. The numbering method takes the data processing unit output by the baseband processing as a unit, and sequentially assigns a natural number number to the data blocks processed by the baseband channel, which is the processing output sequence number. The numbering method adopts circular natural counting, n=1,2,...,N. N is the maximum processing output sequence number, and its value can be set considering the number of available bits and the size of the tag amount. After the count reaches N, it returns to 1 and restarts the numbering.

When writing the baseband processed data into the egress buffer, considering the timeout congestion control process, it will first query the output sequence number table according to the data processing output sequence number to make a judgment to determine whether to write the baseband processed output data to Entry and exit buffers. Wherein, the output sequence number table is used to record the number of the data that has been read from the egress buffer and processed by the front-end, and the specific establishment process will be introduced in step 304 in the read operation of the egress buffer.

When judging whether to write to the export buffer, specifically, read the maximum serial number of the output data from the output serial number table, if the serial number is greater than or equal to the output serial number of the data block to be entered into the export buffer, then judge the data block If it has expired, it does not need to be written into the export buffer, it is discarded directly, otherwise it is judged that the data block has not expired and is written into the export buffer.

Figures 5a and 5b show an example of querying the data output serial number table. In Figure 5a, the processed output sequence number assigned to the baseband processed data block is 4. Compared with the records in the output sequence number table, the maximum sequence number is 2, indicating that the current data block has not expired and can be written into the export buffer. In Figure 5b, the processing output sequence number assigned to the processed data block is 4. Compared with the output sequence number table, the maximum value recorded in it is 5, which means that the data block with the current processing output sequence number 4 has expired and is directly discarded and no longer written. Entry and exit buffers.

Step 304, read data from the export buffer according to the set time interval, record the output serial number of the data in the data output serial number table established, when the export buffer has no data, output empty data, and update the data output serial number table The sequence number of the record.

Since the data output from the baseband processing unit to the front-end has strict real-time requirements, the data blocks sent to the front-end should have strict time intervals. Therefore, in order to ensure that continuous data is sent according to the sending time requirements, it is necessary to monitor whether the serial number counts before and after are continuous. Whether there is data for output after the time interval. specifically,

First, establish an output sequence number table to number and record the data blocks delivered to the front end. The established output sequence number table may be a list of records, each record being the sequence number of a data block that has been delivered to the front end on time, as shown in Table 1. Its initial value is 0, indicating no data output, and the maximum value is the same as the maximum value of the processing output sequence number of the baseband processed data. After reaching the maximum value, the numbering of the subsequent data blocks sent to the front end is restarted, and the previously recorded output sequence number empty.

Output data block sequence number 1 2 3

…

Table 1 Output sequence number table

Then, check whether there is data waiting to be output in the export buffer according to the fixed sending time interval. If there is data, read the data normally, and update the serial number of the output data recorded in the output status table. If there is no data, output empty data for filling, and continue to number and record the empty data in the output serial number table. Empty data refers to data without information content, such as 0.

Figure 6 shows an example of sending data when there is data in the egress buffer, and outputting the record status of the sequence number table. As shown in FIG. 6 , it is assumed that there are data blocks with processing output sequence numbers 4 and 5 waiting to be output in the egress buffer. The data blocks with sequence numbers 2 and 3 have been read from the export buffer, so output sequence numbers 2 and 3 are recorded in the output sequence number table. There is a strict time interval T when data is sent, and within T time after the No. 3 data block is sent, check whether there is data waiting in the export buffer. If there is, send data 4, and mark the No. 4 output in the output sequence number table. In addition, the data blocks that have been processed by the baseband are allocated to the processing output sequence number 6 and written into the buffer.

Give another example of the processing method when timeout congestion occurs. FIG. 7 is an example of sending data and outputting a sequence number table record state when there is no data in the buffer. As shown in Figure 7, due to timeout congestion, there is no processed data output in the baseband. When the buffer still has no data within T time after the No. 3 data is sent, supplementary empty data is sent to the front end. At the same time, it is recorded in the output sequence number table that the fourth block of data has been sent.

It can be seen from the above processing that the output sequence number table will regularly record the output sequence number of the data in the table (including the data actually read from the export buffer and the empty data), that is to say, the output sequence number does not depend on the data in the The processing output number assigned after the baseband processing is the record of the output sequence number that is automatically performed by the output sequence number table. In the case of a normal processing rhythm without overtime congestion, the processing output sequence number assigned to the baseband processed data should be greater than the maximum value recorded in the output sequence number table. This situation shows that the data transmission time after the baseband processing has not been missed. Send real-time requests. When overtime congestion occurs, it may be that the baseband processing is not executed in time, resulting in no corresponding data to be output in the egress buffer, so send empty data at the corresponding sending time, and update the output sequence number table accordingly, waiting for the previously congested data baseband processing When preparing to enter the export buffer zone, it will be found that its transmission time has been missed (i.e. the assigned processing output sequence number of the data after baseband processing described in step 303 is less than or equal to the maximum value in the output sequence number table), then this data does not need to be sent again. Be processed, do not need to enter the export buffer, can be directly discarded. Until it is found that the processing output serial number is greater than the maximum value of the output serial number table, the data is sent to the export buffer for further processing. Through the above processing, the expired baseband processed data can be discarded in time, and no front-end processing is required to ensure the real-time requirement of transmission.

Step 305, when overtime congestion occurs, estimate the overtime congestion degree, and discard outdated data that has not been processed by the baseband.

When there is no data in the buffer, it is usually due to timeout congestion in the baseband processing, which will cause the stagnation of data processing. If the stagnation time is too long, the buffer will have no data output and send empty data. Even if the congested data enters the buffer after subsequent processing, it will not need to be output again because it misses the fixed sending time interval and becomes expired data (that is, the aforementioned steps 303 in the case of discarding data). In addition, if the congestion time is too long, some data that has not entered the baseband processing may also become expired data, and this part of the data can be directly discarded without baseband processing, otherwise there will always be a large delay in subsequent data. The specific treatment is as follows:

First, after the timeout congestion occurs, estimate the timeout congestion degree. When empty data is sent to the front end, it is known that timeout congestion has occurred. Estimate the timeout congestion degree after the timeout congestion ends. The overtime congestion degree is used to measure the amount or time of baseband processing data congestion. It can be determined by the difference between the maximum value in the output sequence number table and the equivalent processing sequence number of the data that is about to enter the baseband when the baseband processed data continues to be output after the timeout congestion. O _M + δ-Ke to represent. Among them, Ke represents the equivalent processing output sequence number of the data that is about to enter the baseband, O _M represents the maximum value in the output sequence number table, and δ is an empirical value, which is set in combination with the baseband processing time and the time interval for sending data to the front end, usually a natural number , to ensure that the data corresponding to O _M +δ+1 can be sent out on time after entering the baseband processing.

Then judge whether the data that has not been processed by the baseband has expired due to congestion according to the timeout congestion degree. If the overtime congestion degree is greater than or equal to 0, it is determined that the corresponding data without baseband processing has expired, and the discarded data without baseband processing is the data corresponding to the equivalent processing output sequence numbers Ke to OM ₊ δ.

An example of a process of judging expired data that has not been processed by the baseband is given below. FIG. 8 is an example of judging expired data that has not been processed by the baseband. Due to the overtime congestion in the baseband processing process, the data that has just been processed by the baseband is allocated to the processing output sequence number 4, and the data with the equivalent processing output sequence number 7 is about to enter the baseband processing unit, but the actual transmission time has been fixed. The No. 8 data is sent to the front end in the form of empty data. At this time, the timeout congestion degree is 8-7=1, which is greater than 0. It is determined that the No. 7 and No. 8 data blocks that have not entered the baseband processing have expired. Combining the baseband processing time and the sending time interval, if δ=1 is set, the data of No. 7-9 will be discarded, and the data of No. 10 and later will be processed by the baseband to meet the real-time requirements sent to the front end.

So far, the flow of the processing method in FIG. 3 ends. In the above process, on the one hand, the baseband processed data is compared and judged to determine whether to discard it; on the other hand, through the judgment of the data before the baseband processing, the data that misses the sending time is directly discarded as soon as possible, and the baseband processing is no longer performed. deal with. The above two aspects of processing are combined with each other to meet the real-time requirements of the sending end and save the processing resources of the baseband resource pool to a greater extent. In fact, for the most basic implementation, the processing of steps 302 and 305 may not be included, and the real-time requirements under certain conditions can also be met; of course, the preferred implementation is the entire processing of Figure 3 including steps 302 and 305 process, so that it can meet the real-time requirements in various situations and save processing resources.

In addition, in the above process, the read operation and write operation to the egress buffer may be performed independently of each other, and may be interspersed with each other in terms of processing time.

Through the process shown in Figure 3 above, for the timeout congestion processing at the baseband processing exit, the baseband processed data is combined with the output sequence number table to determine whether to write to the buffer, and read data from the buffer or fill the empty space at a fixed sending time interval. Sending data, recording the output sequence number, and discarding the data that is not output in time can save processing resources and avoid delays in data transmission caused by congestion, so as to ensure the real-time requirements of the sending end.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the present invention. within the scope of protection.

Claims (1)

1. a downlink channel data processing method in a baseband resource pool, characterized in that, comprising: a. Set the exit buffer at the baseband processing exit in advance; b. Assign an equivalent processing output sequence number to the data before the baseband processing, which is the sequence number of the corresponding data block when the input data of the baseband processing is output after the baseband processing; c. The data after baseband processing is selected and sent to the egress buffer; according to the preset time interval, the data is taken out from the egress buffer for front-end processing, and the data read from the egress buffer is processed by the front-end The processed data is sequentially numbered and recorded; Wherein, when there is no data in the export buffer, read out the empty data and continue to number and record the empty data; The method of sending the data after baseband processing into the export buffer includes: if the processing output sequence number of the data is less than or equal to the maximum number of the currently recorded data for front-end processing, the data is discarded and not sent to the output buffer. the egress buffer; otherwise, send the data into the egress buffer; The processing output sequence number of the data after baseband processing is the sequence number when the data block is output after baseband processing; When the downlink channel has overtime congestion and ends, estimate the overtime congestion degree O _M +δ-K _e , and when the overtime congestion degree is greater than or equal to 0, discard the data with equivalent processing output sequence numbers K _e to O _M +δ, No more baseband processing; Wherein, _OM is the maximum serial number of the currently recorded data that undergoes front-end processing, K _e is the equivalent processing output sequence number of the data that is about to undergo baseband processing, and δ is a preset natural number.