CN101227226B - Method for multi-business scheduling based on communication system real buffer memory - Google Patents

Abstract

A method for multi-service scheduling based on the actual cache volume of a communication system. First, the service priority of each user to be scheduled is not distinguished, but the weight priority of each user to be scheduled is calculated according to the actual cache volume, and the weight priority of each user to be scheduled is calculated according to the weight priority. The sub-channels of the communication system are allocated to each user to be scheduled, and then according to the service priority of each user to be scheduled, it is judged whether all the channels of the communication system have been allocated to the user to be scheduled with a high service priority, if not, then It is further judged whether there are users to be scheduled with high service priority who are waiting to allocate channels, and if so, then the channels corresponding to the users to be scheduled who have been assigned channels with low service priority and low weight priority are reassigned to the on-site Users who are waiting to be allocated channels and have high service priority can enhance the multi-user diversity effect, reduce the filling probability, improve the system throughput, and meet the principle of multi-service processing.

Description

A method for multi-service scheduling based on the actual buffer capacity of the communication system

technical field

The invention relates to a method for multi-service scheduling based on the actual buffering capacity of a communication system. the

Background technique

In recent years, Orthogonal Frequency Division Multiple Access (OFDMA: Orthogonal frequency division multiple access) technology has been adopted by the International Telecommunication Standardization Organization because of its advantages of making full use of various diversity effects and eliminating interference in the cell. The third generation partnership (3GPP) is defined as the downlink air interface technology of the future mobile communication system. the

Orthogonal Frequency Division Multiplexing (OFDM) access technology combines multiple subcarriers into one subchannel and distributes it to one user at a time. Multiple subchannels form a channel set, and multiple users select different subcarriers in the channel set according to channel conditions and service requirements. channel. Obviously, how to allocate sub-channels correctly is the key to the performance of OFDMA systems. Therefore, people pay more and more attention to the scheduling technology specialized in channel allocation. the

The main goal of scheduling technology is to improve the system throughput as much as possible while meeting the user's quality of service (QoS) requirements and fairness. So far, people have proposed many scheduling algorithms for wireless communication, among which the maximum carrier-to-interference ratio (max C/I) algorithm can provide the highest system throughput, but the fairness is the worst; roulette (RR: Round Robin ) algorithm assigns channels to each user in turn, the fairness is the best and the simplest, so it becomes a benchmark algorithm of the scheduling algorithm, but because the wireless channel characteristics and the actual needs of users are not considered, the system throughput is low; Holtzman proposed proportional The fair (PF: Proportional Fair) algorithm has the best overall performance because it takes into account the requirements of both throughput and fairness, so it will soon become the mainstream scheduling technology for wireless communication. At present, researchers have proposed some improved algorithms for some defects of the traditional PF algorithm. For example, Shin S added delay constraints into the PF algorithm to adapt to the transmission of real-time media. Patrick Svedman introduced a fairness adjustment parameter to adapt to different rate requirements of users. Hsiao-ChiangChuang divided users into two types of services: real-time and It is not real-time, and the actual buffer size of the user is used to calculate the waiting delay. If the delay is large, it is allocated first. If it is a non-real-time user, it is necessary to use the traditional PF algorithm for secondary scheduling. the

Since these improved PF scheduling algorithms are mainly academic articles, there are two points that are inconsistent with reality. One is that these algorithms are considered based on sub-carriers, but in practice, scheduling is based on sub-channels. In this way, since the sub-channel is composed of multiple sub-carriers, when the actual buffer size of the user is small, a lot of padding will be generated, resulting in great waste. Therefore, the impact of the user's actual cache volume should be considered in the scheduling algorithm; second, these algorithms do not consider the multi-service environment or the multi-service processing principle does not match the actual situation. It has been clearly defined in the relevant standards of 3GPP that channel allocation must be carried out according to the service priority, that is, after the high priority service is served, the low priority service is served, and the above algorithms do not consider this requirement, so According to Hsiao-Chiang Chuang's algorithm, there will be a phenomenon that the FTP service will occupy the VOIP service, which is intolerable in practice.

Therefore, in practice, people usually still use the traditional PF algorithm to schedule high-priority services first, and then schedule low-priority services. In this way, when some users with high service priority but small data volume coexist with some users with low service priority but large data volume, such as VOIP users and FTP users, the actual buffer size will not be considered. It appears that VOIP users occupy good channels first, but for VOIP users, it is extremely wasteful to occupy too good channels, because they do not have too much data to transmit, and FTP users with a large amount of data can only occupy poor channels , which will reduce the throughput of the system. the

To sum up, how to solve many shortcomings in the prior art has become a technical problem to be solved urgently by those skilled in the art. the

Contents of the invention

The purpose of the present invention is to provide a method for multi-service scheduling based on the actual buffer amount of the communication system, so as to enhance the multi-user diversity effect, reduce the filling probability, realize the improvement of system throughput, and meet the multi-service processing principle at the same time. the

In order to achieve the above object, the present invention provides a method for multi-service scheduling based on the actual buffer amount of the communication system, which includes: 1) not distinguishing the service priority of each user to be scheduled, but only based on the actual buffer amount of each user to be scheduled Calculate the weight priority of each user to be scheduled; 2) assign each channel of the communication system to the corresponding user to be scheduled according to the weight priority of each user to be scheduled; 3) according to the weight priority of each user to be scheduled Scheduling the user's service priority to determine whether all channels of the communication system have been allocated to users to be scheduled with high service priority; 4) when not all channels of the communication system are allocated to users to be scheduled with high service priority 5) If there are users to be scheduled with high service priority waiting to be allocated channels, the assigned users with low service priority and low weight priority The channel corresponding to the user to be scheduled is re-assigned to the user to be scheduled who is waiting to be allocated a channel and has a high service priority. the

计算各待调度用户的权重优先级，其中 ，且当DRC_i(t)≤R_buffer，i(t)时，DRC_i(t)＝DRC_i(t)，当DRC_i(t)＞R_buffer，i(t)时，DRC_i(t)＝R_buffer，i(t)，DRC_i(t)为待调度用户i的信道速率，R_buffer，i(t)为所述待调度用户i的实际缓存量， 为所述 待调度用户i的平均速率，该平均速率以t_c长的时间窗进行更新。 Wherein, in the step 1) according to

Calculate the weight priority of each user to be scheduled, where , and when DRC _i (t)≤R _{buffer, i} (t), DRC _i (t)=DRC _i (t), when DRC _i (t)>R _{buffer, i} (t), DRC _i (t )=R _{buffer, i} (t), DRC _i (t) is the channel rate of the user i to be scheduled, R _{buffer, i} (t) is the actual buffer amount of the user i to be scheduled, is the average rate of the user i to be scheduled, and the average rate is updated with a long time window of _tc .

To sum up, the method for multi-service scheduling based on the actual buffer capacity of the communication system of the present invention firstly performs channel allocation for the first time according to the weight priority of each user to be scheduled, and then allocates the channels that have not been scheduled in the first channel allocation. The users to be scheduled with high service priority will be assigned a second channel according to their service priority, which can enhance the multi-user diversity effect, reduce the filling probability, improve the system throughput, and meet the requirements of multi-service processing in principle. the

Description of drawings

FIG. 1 is a schematic diagram of the operation flow of the method for multi-service scheduling based on the actual buffer capacity of the communication system according to the present invention. the

Figure 2 is a schematic diagram of system throughput comparison. the

Fig. 3 is a schematic diagram of comparing channel allocation fairness of FTP users. the

Detailed ways

Please refer to Fig. 1, in the method for multi-service scheduling based on the actual buffering amount of the communication system of the present invention, step S10 is first performed, and the service priority of each user to be scheduled is not distinguished, and only the actual buffering amount of each user to be scheduled is calculated. Describe the weight priority of each user to be scheduled, which is based on the formula $P_i\left(t\right)=\frac{{\mathrm{DRC}}_i\left(t\right)}{\stackrel{\‾}{R_i\left(t\right)}}$ Calculate the weight priority of each user to be scheduled, where $\stackrel{\‾}{R_i\left(t\right)}=(1-\frac{1}{t_c})*\stackrel{\‾}{R_i(t-1)}+\frac{1}{t_c}*{\mathrm{DRC}}_i(t-1),$ And when DRC _i (t)≤R _{buffer, i} (t), DRC _i (t)=DRC _i (t), when DRC _i (t)>R _{buffer, i} (t), DRC _i (t) =R _{buffer, i} (t), DRC _i (t) is the channel rate of user i to be scheduled, R _{buffer, i} (t) is the actual buffer amount of user i to be scheduled, R _i (t) is the The average rate of user i to be scheduled, the average rate is updated with a time window of _tc length, for example, if there are users A and B to be scheduled in the communication system, and the communication system has two channels x1 and x2, if Scheduling user A occupies the x1 channel, and its weight priority a1 is calculated as 1. If the user A to be scheduled occupies the x2 channel, its weight priority a2 is calculated as 3. If the user B to be scheduled occupies the x1 channel, its weight priority is calculated. The level b1 is 2. If the user B to be scheduled occupies the x2 channel, its weight priority b2 is calculated as 2, and then step S11 is executed.

In step S11, according to the weight priority of each user to be scheduled, the PF algorithm is used to allocate each channel of the communication system to the corresponding user to be scheduled, for example, according to the calculated Weight priority, allocate x1 channel to user B to be scheduled, allocate channel x2 to user A to be scheduled, and then perform step S12. the

In step S12, judge whether all the channels of the communication system have been allocated to users to be scheduled with high service priority according to the service priority of each user to be scheduled, if it is finished, otherwise perform step S13, in this embodiment, User B to be scheduled is an FTP user whose service priority is low. the

In step S13, it is judged whether there are users to be scheduled with high service priority waiting to be allocated channels, if so, execute step S14, otherwise, end. For example, user c to be scheduled in the communication system is a VoIP user, and obviously the service priority is high. the

In step S14, the channel corresponding to the user to be scheduled who has been assigned a channel with a low service priority and a low weight priority is reallocated to the user to be scheduled who is waiting to be allocated a channel and has a high service priority. The x1 channel assigned to user B to be scheduled is reassigned to user c to be scheduled. In addition, when there are multiple users to be scheduled who have already allocated channels, their service priorities are lower than those of users who are waiting to be allocated channels and have high service priority. For users, usually the channels corresponding to the scheduled users with the lowest weight priority and assigned sub-channels are preferentially assigned to users with high service levels who are not assigned sub-channels, and then return to step S12 for a new round of judgment. the

The superior performance of the method for multi-service scheduling based on the actual buffer amount of the communication system of the present invention will be further described below. First, it is assumed that there are 30 users, and they are divided into two kinds of services: VOIP and FTP. Will adopt the present invention and existing two kinds of commonly used scheduling methods namely RR (VOIP)+PF (FTP) and PF (VOIP)+PF (FTP) to compare, wherein, RR (VOIP)+PF (FTP) refers to First use the RR algorithm to schedule VOIP users and then use the PF algorithm to schedule FTP users, while PF(VOIP)+PF(FTP) first use the PF algorithm to schedule VOIP users and then use the PF algorithm to schedule FTP users. Both algorithms are scheduled first. After the VOIP user, the FTP user is scheduled again. The simulation conditions are shown in Table 1:

Table 1 is the simulation conditions

parameters value system bandwidth 5 MHz cell diameter 1500 meters Scenes Suburban macro User number 30 The number of subcarriers contained in each subchannel 25 Scheduling time (TTI) 0.5 milliseconds business model VOIP, FTP Modulation and coding method Four-phase keying (QPSK) (R＝1/3, 1/2, 2/3, 3/4, 4/5), 16QAM (R＝1/2, 2/3, 3/4, 4/5 ) Hybrid Request Retransmission (HARQ) none thermal noise spectral density -174dBm/Hz Simulation time 20000TTI

Please refer to Figure 2 and Figure 3, wherein Figure 2 is a schematic diagram of system throughput comparison, the horizontal axis is the number of VOIP users, the maximum is 9 VOIP users, and the total number of users is 30, so the number of FTP users is 21 at this time. The vertical axis is throughput in kilobits per second. It can be seen from the figure that the throughput of the present invention is higher than that of the existing two methods in any case. Fig. 3 is a schematic diagram of comparing channel allocation fairness of FTP users, which represents whether each FTP user has an equal opportunity to allocate channels. The horizontal axis is the number of FTP users, and the vertical axis is the fairness of channel allocation, the highest being 1. As can be seen from the figure, the present invention is also slightly better than the existing two methods in terms of fairness. the

To sum up, the method for multi-service scheduling based on the actual buffer capacity of the communication system in the present invention improves the multi-user diversity effect and reduces the filling probability, while ensuring user fairness, improves the system throughput, and also satisfies the requirements of multiple users. Principles of business processing. the

Claims (1)

1. one kind is carried out the method for multi-business scheduling based on communication system real buffer memory, it is characterized in that comprising step:

1) only calculates the described weight priority of respectively treating dispatched users based on the real buffer memory of respectively treating dispatched users;

2) adopt the direct ratio fair algorithm that each channel allocation of described communication system is given according to the described weight priority of respectively treating dispatched users and treat dispatched users accordingly;

3) according to the service priority of respectively treating dispatched users judge all channels of described communication system whether all distributed to service priority high treat dispatched users;

4) when all channels of described communication system be not all distribute to service priority high treat dispatched users the time, judge whether that the dispatched users for the treatment of that service priority is high in addition waiting for allocated channel;

5) if there is the high dispatched users for the treatment of of service priority waiting for allocated channel, with service priority low and weight priority is low be assigned channel treat the pairing channel of dispatched users redistribute to described wait for allocated channel and service priority high treat dispatched users;

Described step 1) basis

Calculate the priority of respectively treating dispatched users, wherein

And work as DRC _i(t)≤R _{Buffer, i}(t) time, DRC _i(t)=DRC _i(t), work as DRC _i(t)＞R _{Buffer, i}(t) time, DRC _i(t)=R _{Buffer, i}(t), DRC _i(t) for treating the channel speed of dispatched users i, R _{Buffer, i}(t) be the described real buffer memory for the treatment of dispatched users i,

Be the described Mean Speed for the treatment of dispatched users i, this Mean Speed is with t _cLong time window upgrades.

Images