CN104270723A - Broadcast Scheduling Method for Multi-channel Big Data Items Based on Two-layer Scheduling Strategy - Google Patents

Abstract

The invention discloses a multi-channel large data item broadcast scheduling method based on a double-layer scheduling strategy. The method includes three processes including large data item subpackaging, high-level data item channel allocating and low-level single channel data item allocating. In the initial stage, a server side judges whether large data items are included according to channel capacity of available channels and the data length of all required data items. If the large data items are included, a pre-subpackaging mechanism needs to be carried out and subpackaging is reasonably carried out. A high-level data item channel allocating algorithm is adopted, and the algorithm guarantees optimal channel allocation; meanwhile, the algorithm is optimized, and accordingly the situation that the channels skip many times when mobile subscribers obtain multiple data items is avoided, and average access time is reduced. In actual application, a feedback mechanism can be introduced for a low-layer single channel data item scheduling strategy, when the subscribers of mobile terminals receive data, the server side is fed back, if certain data are not required, the data can be removed in the next pushing cycle, and the average access time is greatly reduced.

Description

Broadcast Scheduling Method for Multi-channel Big Data Items Based on Two-layer Scheduling Strategy

technical field

The invention belongs to data transmission communication technology, and in particular relates to a multi-channel large data item broadcast scheduling method based on a two-layer scheduling strategy.

Background technique

Broadcasting data through wireless channels can not only overcome the asymmetry and low bandwidth disadvantages of mobile wireless communication, but also support a large number of mobile devices to efficiently access the server-side database at the same time, and the cost of sending data is basically independent of the number of receivers. Due to the limited energy consumption of mobile devices, data broadcasting can reduce the power consumption of acquiring data. Wireless data broadcasting is currently the mainstream method of wireless network data transmission. The broadcast server sends data through the public channel, and the mobile terminal users listen to the broadcast channel and download the data they need in time.

How to quickly obtain and access data and reduce the power consumption of mobile devices are two key points in the research of wireless data broadcasting. Correspondingly, the two main parameters to measure broadcast performance are: access time (Access time) and tuning time (Tuning time). Access time refers to the time consumed from the user making a request to obtaining the requested data. Tuning time refers to the time spent listening to the channel from when the user makes a request to when the requested data is obtained.

Currently, the mainstream broadcast scheduling algorithms include flat broadcast (Flat Broadcast) and access probability-based scheduling (Probabilistic based Broadcast) for single data items (only one data item is requested each time). For multiple data item requests, there are QEM scheduling algorithm, improved QEM scheduling algorithm, GCM algorithm, etc. For each scheduling algorithm of multiple data items, the QEM algorithm is the first broadcast algorithm to discuss multiple data items in detail. Compared with the random scheduling algorithm, the average access time is reduced by about 25%; the improved QEM algorithm averages The access time is shorter than the QEM algorithm, but the complexity is higher; the average access time of the GCM algorithm is longer than the QEM algorithm in all cases, but the complexity is lower.

Multi-channel wireless data broadcasting refers to simultaneous data broadcasting on multiple channels. Compared with single-channel data broadcasting, efficiency is improved and access time is reduced. Although in recent years, many researchers have proposed some data broadcast scheduling algorithms for multi-channel single data item requests, such as two-level optimization scheduling algorithm (TOSA). However, the scheduling of multiple data item requests under multiple channels is still in the experimental stage. For a large data item with a long data length, if the data item is directly regarded as a whole, it will cause the large data item to be pushed during the channel. If other users demand the data item and start listening to the channel, You need to wait until the end of this data push cycle and start to push the data item in the next cycle before you can start receiving it. Not only the tuning time is greatly increased, but also the access time.

Contents of the invention

The purpose of the present invention is to propose a multi-channel large data item broadcast based on a two-layer scheduling strategy for how to efficiently and quickly obtain multiple data items containing large data items for mobile devices with limited energy consumption in the case of asymmetric wireless networks Scheduling method.

To achieve the above object, design of the present invention is:

A channel allocation algorithm is used to allocate multiple data packets to channels after subpacket preprocessing for large data items with long data item lengths. Realized that under the push of big data items, when other users need real-time data during the periodic broadcast data period, the subcontracting application enables users to get the data without waiting for a long time, reducing the average access time. At the same time, the request of multi-channel and multi-data items is realized, and the average access time is reduced by reducing data access conflicts and channel jumps.

According to above-mentioned inventive concept, the present invention adopts following technical scheme:

A multi-channel big data item broadcast scheduling method based on a two-layer scheduling strategy, including three processes of big data item data packetization, high-level data item channel allocation, and low-level single-channel data item allocation. The specific steps are as follows:

a. Data subcontracting of large data items, including the following steps:

(a-1), initialization, the server obtains the unique identification of all mobile devices requesting data, requiring each mobile device to have an independent identification number;

(a-2), the total number of available channels obtained by the server is counted as K, and all channels are numbered, requiring each channel to have an independent number;

(a-3), the server obtains all data items accessed by mobile users, the total number is N, and numbers all data items, requiring each data item to have a unique number;

(a-4), according to the number of each channel in step (a-2), obtain the corresponding available channel bandwidth as ,in ;

(a-5), according to the number of each data item in step (a-3), count the access probability of each data item as p, and the data packet length of each data item as l;

(a-6), according to the bandwidth of each available channel and the data packet length of each data item, determine whether the required data item contains a large data item, if it contains, it needs to pre-packet the large data item, if not, directly according to The channel allocation method allocates channels to each data item;

(a-7), according to the bandwidth of the largest available channel, pre-package large data items, and number all data packets, requiring each data packet of the data item to have a unique identifier, and the number contains the data Item identification information;

(a-8), select a channel with a larger available channel bandwidth, and reserve the channel;

b. High-level data item channel allocation, including the following steps:

(b-1), select the reserved channel with the largest available channel bandwidth, and allocate each data packet divided by the big data item to the channel;

(b-2) According to the product of the access probability of each data item and its data length, divide the remaining data items except the allocated large data items, and arrange the data items according to the obtained product from large to small ;

(b-3), according to the channel bandwidth of the other channels except the channel that has been allocated, and arrange the channels according to the bandwidth of the channel from small to large, record the total number of remaining available channels as ;

(b-4), order ,in Indicates the channel bandwidth of the i-th available channel, Indicates the first The channel bandwidth of available channels;

(b-5), each sorted 2B data items are taken as a group, and the first B data items are assigned to 1 to , the last B data items are sequentially assigned to to 1, the number of data items allocated to each channel is proportional to the square root of the channel's available bandwidth;

(b-6), order It is equal to the sum of the square root of the product of the access frequency and the length of all data items assigned to the i-th channel, and recursively compares each channel in turn The value of , find the channel with the largest and smallest value and calculate it as ;

(b-7), from Call each data item recursively in turn to find the value The smallest data item counts as ;

(b-8), pair channel ,Compare and The size of the value, if the former is greater than the latter, the channel The smallest data item in move to channel , if the former is less than or equal to the latter, keep the original state;

(b-9), complete the high-level data item channel allocation algorithm;

c. Allocation of low-level single-channel data items, including the following steps:

(c-1), initialization stage, remove the channel for transmitting large data items, for any allocated data channel i, take the current time T=0, let Calculate the access frequency of all data items on the channel respectively as and the length is denoted as ,data item The interval is , where M is the sum of allocated data items for channel i and ;

(c-2), stipulate that the transmission sequence set of each channel data item is ;

(c-3), compare the data items of each channel The value of , find the data item with the smallest value and record it as , and ;

(c-4), each time a data item is added to the set S, let represents the current moment;

(c-5), the current time T, let ,satisfy and The smallest data item is put into the data set S;

(c-6). Recursively recursively the above algorithm for all data items on the channel until all data items allocated on the channel are scheduled.

Compared with the prior art, the present invention has the following obvious outstanding substantive features and significant technical progress:

The method of the present invention introduces the idea that multiple data items contain large data items, and according to the maximum channel capacity, the idea of pre-packetizing and then sub-packetizing and pushing reduces the average access time of data items, and at the same time reduces the number of channel jumps and data access conflicts. The power consumption of the mobile terminal is reduced. The invention absorbs the advantages of the two-layer scheduling strategy, and improves the algorithm (including the channel allocation algorithm and the log-time algorithm) of the multi-channel single data item request two-layer scheduling strategy. By adopting the method of the present invention, it can be realized that other mobile devices need to obtain the data item in real time during the period of pushing the big data item, without waiting for too long, and the average access time is reduced.

Description of drawings

Figure 1 is a structural diagram of the data broadcasting system.

Fig. 2 is a schematic diagram of prepacketization processing in the present invention.

Fig. 3 is a flow chart of big data prepacketization processing in the present invention.

Fig. 4 is a flowchart of high-level data item channel allocation in the present invention.

Fig. 5 is a schematic diagram of the results of the single-channel data item allocation algorithm of the present invention.

Fig. 6 is a flow chart of the single channel data item allocation algorithm of the present invention.

Detailed ways

Preferred embodiments of the present invention are described in detail as follows in conjunction with accompanying drawings:

A multi-channel big data item broadcast scheduling method based on a two-layer scheduling strategy, including three processes of big data item data packetization, high-level data item channel allocation, and low-level single-channel data item allocation. The specific steps are as follows:

a. Data subcontracting of large data items, including the following steps:

(a-1), initialization, the server obtains the unique identification of all mobile devices requesting data, requiring each mobile device to have an independent identification number;

(a-2), the total number of available channels obtained by the server is counted as K=3, and all channels are numbered, each channel is required to have an independent number, which is counted as , , ;

(a-3). The server obtains all the data items accessed by mobile users. The total count is N=7, and all data items are numbered. Each data item is required to have a unique number, which is counted as , , , , , , ;

(a-4), according to the number of each channel in step (a-2), obtain the corresponding channel bandwidth and count as =10, =10, =20, such as =20 means channel Data items with a length of 20 units can be transmitted;

(a-5), according to the number of each data item in step (a-3), the access probability of each data item is calculated as =0.5, =0.2, =0.1, =0.1, =0.07, =0.05, =0.01, and the packet length of each data item is calculated as =1, =16, =3, =3, =2, =3, =2, such as Represents a data item The length of is 2 unit lengths;

(a-6), according to the bandwidth of each available channel and the data packet length of each data item, determine whether the required data item contains a large data item, by ,Sure is a big data item,;

(a-7), based on the bandwidth of the largest available channel =20, for large data items Perform pre-packet processing, and number all data packets, will Each four-unit data length is used as a data packet for grouping, and the data packets are counted as , , , ;

(a-8), reserved channel for transferring data items Each data packet of

b. High-level data item channel allocation, including the following steps:

(b-1), select the channel , the big data item Divided data packets , , , assigned to the channel;

(b-2), according to each data item The value of , dividing the allocated large data item The rest of the data items, and according to the product obtained from small to large, the data items are arranged in order as , , , , , ;

(b-3), remove the channel , and the remaining channels are arranged in ascending order of bandwidth as follows: , , the total number of available channels after sorting is counted as ;

(b-4), pursuant to step (b-3) order 2, of which Indicates the i-th ( ) the channel bandwidth of available channels, Indicates the maximum value of the channel bandwidth in (b-3);

(b-5). Each 2B=4 data items that have been sorted are taken as a group. On the premise of ensuring that the bandwidth of each channel is sufficient, the first 2 data items are sequentially allocated to 1 to , the last 2 data items are sequentially assigned to to 1, the result is , , assigned to , , , assigned to ;

(b-6), order is equal to assigning to the i-th The sum of the quadratic root of the product of the access frequency and the length of all data items of a channel, recursively compare each channel worthy of ,remember ;

(b-7), from Call each data item recursively in turn to find the value The smallest data item counts as ;

(b-8), pair channel ,Compare and The magnitude of the value results in: , there is no need to put the data item move to channel ;

(b-9), complete the high-level data item channel allocation algorithm, and the result is the data item , , assigned to channel, data item , , assigned to channel;

c. Allocation of low-level single-channel data items, including the following steps:

(c-1), initialization phase, The channels respectively take the current time T=0, , respectively calculate each channel data item The interval is , where M is the sum of data items assigned to each channel i and . channel: , , , channel: , , ;

(c-2), stipulate that the transmission sequence set of each channel data item is , the current moment, Channel S sets are all sets;

(c-3), compare each data item of each channel separately The value of , find the data item with the smallest value and record it as , and , the current moment channel , channel ;

(c-4), Channel only remaining data items but The data transmission scheduling sequence of the channel is , channel current moment ;

(c-5), Channel current time T=1, let ,but , ,satisfy and , then add data item in S data set ;

(c-6), Channel only remaining data items but The data transmission scheduling sequence of the channel is ;

(c-7), to sum up The channel S dataset is , The channel S dataset is .

In the actual system design process, there are usually many mobile terminal users who need to access one or more same data items. Referring to Figure 1, the architecture diagram of the data broadcasting system, both the information server and the broadcasting server can use a wired network to connect to the database to directly retrieve data, and the mobile terminal needs to listen to the wireless channel through which the broadcasting server pushes data through the wireless network. In practice, many mobile terminals need to obtain a large number of relevant user information based on the geographic location. If all the data is regarded as a data item as a whole, when the data item is broadcast, other mobile terminals that need the data item will start to monitor the channel. It takes a long time to wait. If the large data item is prepackaged, other users can start receiving part of the data first. The comparison process is shown in Figure 2, a schematic diagram of pre-packetization processing.

In the process of actually pushing data, multiple mobile terminals need to monitor the same available channel in order to obtain the same data, and different mobile users will access the channel at different times. The flow chart of big data pre-packetization processing is shown in Figure 3. First, the server obtains the channel capacity of each available channel, such as the maximum number of data packets allowed to be transmitted by the channel, and the maximum number of data items allowed to be transmitted. Second, the server gets the length of all requested data items. Based on the channel capacity of all available channels and the length of the data item, it is judged whether there is a large data item. If there are large data items, it is necessary to select the channel with the largest channel capacity, determine the packetization strategy according to its channel capacity, and pre-package the large data items. To number each data packet, each data packet is required to have a unique identifier, and the number contains the identifier of the data item. Then sort by package serial number.

The high-level data item channel allocation flow chart is shown in Figure 4. First, the server side judges whether there is a large data item. If it does not exist, it will directly execute the allocation algorithm of the available channel data items; if it exists, the server side needs to judge whether the pre-packet processing of large data items has been completed. If it has been completed, it is necessary to allocate the data packets of the large data items to the channel with the largest reserved channel capacity according to the sorted order, and then execute the channel allocation algorithm; if it is not yet completed, wait for the mechanism to complete. After the allocation algorithm of the available channel data items is completed, the server needs to optimize the algorithm. The server needs to count all the data items accessed by different users, and fine-tune the allocated data items according to the user classification, so that the data items required by the same user More allocation to the same channel avoids the need for channel jumping because the accessed data items are on different channels, and increases the average access time of the mobile terminal.

After the high-level data item allocation algorithm is executed, the single-channel data item allocation algorithm is finally executed on each channel. The result diagram of this algorithm is shown in Figure 5. In practical applications, different mobile terminals will access the channel at different times, so the data items with high access frequency should be evenly distributed on the channel. When other mobile users need to obtain When the data item is received, the data item can be received without waiting for the end of the broadcast period, which saves the average access time and reduces the energy consumption of the mobile terminal. The algorithm flow chart is shown in Figure 6. In practical applications, the server first needs to enter the initialization phase to obtain the length and access frequency of all data items allocated on the channel, and specify the data sets that meet the conditions according to the actual algorithm requirements. Each time the recursive log-time algorithm adds the data items satisfying the condition to the data set one by one. It is judged whether all the channels that allocate data items need to complete the algorithm, and if it is completed, the two-layer data item scheduling algorithm can end. If not, all unfinished channels need to be completed.

After the execution of the multi-channel big data item broadcast scheduling algorithm based on the two-layer scheduling strategy is completed, under the background conditions of multiple mobile terminal users accessing multiple data items, when the required data item contains big data, pre-package through big data The processing can reduce the average waiting time of mobile terminal users, and the optimization of the channel allocation algorithm of upper layer data items avoids multiple channel jumps when each mobile user obtains multiple data items, and also reduces the average access time. In practical applications, the low-level single-channel data item scheduling strategy can introduce a feedback mechanism. When each mobile terminal user receives the data, it will give the server feedback. If a certain item of data is no longer needed, the data can be removed in the next push cycle, greatly Reduced average access time. the

Claims (1)

1. based on the large data item broadcast dispatching method of multichannel of double-deck scheduling strategy, it is characterized in that, comprise large data item data subpackage, high level data item channel allocation, low layer single channel data item distribute three processes, concrete steps are as follows:

A, large data item data subpackage, comprise the steps:

(a-1), initialization, server end obtains the unique identification of the mobile device of all request msgs, requires that each mobile device has an independently identify label number;

(a-2), server end obtains the sum of available channel and counts K, and be numbered all channels, requires that each channel has one and independently numbers;

(a-3), server end obtains the data item of all mobile subscribers access, and sum counts N, and to all data item incomes numbering, requires that each data item has a unique numbering;

(a-4), according to the numbering of step (a-2) each channel, obtain corresponding available channel bandwidth and count , wherein ;

(a-5), according to the numbering of step (a-3) each data item, the accessed probability that statistics obtains each data item counts p, and the data packet length of each data item counts l;

(a-6), according to the data packet length of each data item of bandwidth sum of each available channel, whether judge in demand data item containing large data item, if containing, need to carry out the pre-subpackage process of large data item, if without, direct basis channel allocation method carries out channel allocation to each data item;

(a-7), according to the bandwidth of maximum available channel, pre-subpackage process is carried out to large data item, and all packets are numbered, require that each packet of this data item has a unique mark, and the identification information of numbering containing this data item;

(a-8), the larger channel of available channel bandwidth is chosen, and this channel reserved;

B, high level data item channel allocation, comprise the steps:

(b-1), the reserved channel with maximum available channel bandwidth is chosen, by each allocation of packets ready-portioned for large data item to this channel;

(b-2), according to the accessed probability of each data item and the product of its data length, divide the remainder data item except the large data item distributed, and arrange each data item from big to small according to the product obtained;

(b-3), according to the channel width of all the other channels except the channel distributed except, and arrange each channel from small to large according to the bandwidth of channel, note remains available channel and adds up to ;

(b-4), make , wherein represent the channel width of i-th available channel, represent the the channel width of individual available channel;

(b-5), using every 2B data item having sorted as one group, front B data item order-assigned to 1 to in, rear B data item order-assigned arrives in 1, the quantity square root that channel available is wide being therewith assigned to the data item of each channel is directly proportional;

(b-6), make equal to be dispensed to the accessed frequency of all data item of i-th channel and the summation of length product square root, successively the more each channel of recurrence value, find the minimum and maximum channel of this value and count ;

(b-7), from in each data item of recursive call successively, find value minimum data item is counted ;

(b-8), to channel , compare with the size of value, if the former is greater than latter by channel in minimum data item move to channel if the former is less than or equal to the latter, then maintain the original state;

(b-9) high level data item channel allocation algorithm, is completed;

C, low layer single channel data item are distributed, and comprise the steps:

(c-1), initialization rank, remove the channel of the large data item of transmission, to any distribute data channel i, get current time T=0, order calculate the accessed frequency of all data item on this channel to be respectively designated as be designated as with length , data item be spaced apart , wherein M be channel i distribute data item summation and ;

(c-2), specify that the set of each channel data item transmission sequence is ;

(c-3), each data item of more each channel value, find the minimum data item of this value and be designated as , and ;

(c-4), after each S set added a data item, made represent current time;

(c-5), current time T, order , meet and that minimum data item puts into data set S;

(c-6), to data item all on this channel successively above-mentioned algorithm of recurrence, until all data item that this channel is assigned with all have been dispatched.