KR20030078543A - Method for scheduling down links based on user utility in wireless telecommunication system - Google Patents

Abstract

PURPOSE: A method for scheduling in a downlink according to user satisfaction in a wireless communication system is provided to assign slots by selecting a user having the maximum satisfaction sum every scheduling period, thereby minimizing service delay time and improving substantial satisfaction. CONSTITUTION: Whenever a scheduling period starts(S105), a scheduler of a base station connects communication with the base station, and decides whether a user wanting to receive data exists(S110). If not, a current slot is maintained in idle state. If at least one user wanting to receive the data exists(S115), the current slot is assigned to the user(S120). The base station transmits the requested data to the at least one user from the current slot. If more than two users exist, a scheduling step starts in accordance with steps(S125-S155).

Description

Downlink scheduling method based on user satisfaction in wireless communication system {METHOD FOR SCHEDULING DOWN LINKS BASED ON USER UTILITY IN WIRELESS TELECOMMUNICATION SYSTEM}

The present invention relates to a wireless cellular data system, and more particularly to a method for scheduling transmission of data packets on the downlink.

In general, wireless communication systems such as Code Division Multiple Access (CDMA) 2000, Wideband Code Division Multiple Access / Universal Mobile Telecommunications System (WCDMA / UMTS), General Packet Radio System (GPRS), and CDMA2000 1xEV-DO (Evolution Data Only) The third generation wireless communication system. The third generation wireless communication system supports a high speed packet data service as well as a voice service, unlike a typical second generation wireless communication system which only supports a voice service or a low speed data service.

In a wireless communication system, a code division multiplex (CDM) or a time division multiplex (TDM) scheme may be used to transmit data packets to a plurality of users, or both schemes may be used together. In the case of a voice service in a wireless communication system, one code or time slot may be allocated to a user, but the number of time slots required for a data service varies according to the size of a packet to be serviced for each user, and the transmission rate is also variable. Therefore, the efficient allocation of resources in data services is a very important issue.

1 is a diagram illustrating code division multiplexing and time division multiplexing in a wireless communication system.

As shown in (a) of FIG. 1, code division multiplexing can assign a code to each user and serve multiple people at the same time. However, if the relationship between codes does not maintain complete independence, interference occurs and codes cannot be flexibly allocated according to the size of a packet to be transmitted. On the other hand, the time division multiplexing shown in FIG. 1B can be used by only one assigned user. However, when the packet size is large, a large number of time slots can be used. There is no advantage to this.

For data services that are not significantly affected by the continuity of transmission, time division is known to be more efficient. As mentioned earlier, in a time division, only one assigned user can use one time slot, so when multiple terminals are connected to the base station, selecting a user to use each time slot can increase the efficiency of data transmission. It is a very important task. This task is called scheduling.

The wireless communication system has an uplink for transmitting data to a base station and a downlink for transmitting data to a terminal. The uplink and the downlink are distinguished by using different codes according to the code division multiplexing scheme. In the data service, the amount of data packets transmitted through the downlink rather than the uplink is largely large. Therefore, various techniques for efficient scheduling in downlink have been studied.

As an example of a known scheduling technique, F. Kelly, A. Maulloo and D. Tan, "Charging and rate control for elastic traffic", European Transactions on Telecommunications, Vol 8, pp 33-37, 1997, describe proportional fairness. The present invention proposes a scheduling scheme based on a) and is applied to a high-speed data rate (HDR). If a user's data throughput is increased by x% than when using a particular scheduling algorithm, then the reduction in data throughput that all other users will experience is greater than x%. It can be said that it is used as a standard of fairness, and in this case, scheduling is performed to satisfy Equation 1 below.

Here, i is an identifier for identifying a user in the set S of users connecting data communication with the base station. Is the transfer rate given to each user according to the proportional fairness criterion Is the possible data rate of each user when using any other scheduling scheme. In this scheduling scheme, scheduling is performed only by considering a user's transmission rate. However, this does not consider the priority of the service according to the service grade and traffic characteristics of each user.

As another example of a known scheduling technique, A. Jalali, R. Padovani and R. Pankaj, "Data Throughput of CDMA-HDR a High Efficiency High Data Rate Personal Communication Wireless System", IEEE VTC 2000, Discuss the scheduling algorithm used in the system. At any arbitrary time t, where t is the slot index, the rate required by the user i to the base station And the average transfer rate of user i In this case, this scheduling algorithm considers the following conditions.

1) of all users with data to be sent at a certain scheduling time t Of The slot is allocated by selecting the user with the highest value. If there are users with the same value, any one of them is selected.

2) Average transfer rate of user i if slot is allocated Is updated as shown in Equation 2 below.

Where above Silver weights Average transfer rate obtained using. In other words, The larger the past bitrate Is reflected at a high rate The smaller the current transmission rate This is reflected at a high rate. Also above If the value is set to the current requested transmission rate for the user who is assigned the current slot and 0 for the remaining users, the average transmission rate can be calculated for users who have no data to transmit. However, this scheduling simply uses the average rate based on the past rate and does not adapt to the change of channel state, which makes it weak for the user's mobility.

As another example of a known scheduling technique, Xin Liu, Ediwin K.P.Chong and Ness B. Shroff, "Transmission Scheduling for Efficient Wireless Utilization", IEEE INFOCOM 2001, consider a utility function in terms of each user. Unlike using the same satisfaction function in the above-described scheduling schemes, this scheduling scheme allows different users to set different satisfaction functions according to the channel state of the user, that is, the signal to noise ratio (SNR).

That is, the user satisfaction function is set differently for each user according to the power noise ratio (SNR), and scheduled to maximize the sum of the satisfaction functions of all users. Although this scheduling scheme sets up the satisfaction function of each user differently, it only takes into account the user's channel status, so it does not consider the length or delay of data packets, which greatly affects the user's satisfaction. there was.

Accordingly, an object of the present invention, which was devised to solve the problems of the prior art operating as described above, is to provide a method for scheduling data packets in a wireless data communication system.

Another object of the present invention is to provide a method for setting user satisfaction in consideration of the length of a data packet.

Another object of the present invention is to provide a method for setting user satisfaction in consideration of delay time.

It is still another object of the present invention to provide a method for scheduling data packets according to satisfaction set uniquely for each user.

An embodiment of the present invention, which was created to achieve the above object, is a wireless communication system that supports time division multiplexing to transmit data packets of a desired size to a plurality of users who are wirelessly communicating with a base station. In the downlink scheduling method,

Setting different satisfaction levels for the plurality of users;

Selecting a user who maximizes the sum of satisfaction of all users among the plurality of users by using the set satisfaction level in every scheduling period;

Allocating a timeslot for data transmission to the selected user.

1 illustrates code division multiplexing and time division multiplexing in a wireless communication system.

2 illustrates a wireless communication network model to which the present invention is applied.

3 is a diagram illustrating a network configuration of a wireless communication system to which the present invention is applied.

4 is a diagram showing the configuration of a base station (BTS) shown in FIG.

5 is a view showing the configuration of the channel cards shown in FIG.

6 is a diagram showing a satisfaction function set for each user according to the present invention in the form of a curve.

FIG. 7 shows a comparison of continuous scheduling and discontinuous scheduling. FIG.

8 is a flowchart illustrating a downlink scheduling operation according to the present invention.

FIG. 9 is a graph illustrating the scheduling algorithm according to the present invention in FIG. 6 compared with the prior art using proportional fairness without considering delay time. FIG.

10 and 11 are graphs showing the cumulative density function and the overall yield of the average delay time when the user moves by comparing the present invention with the prior art.

12 is a diagram showing a satisfaction function used in the case of applying a different satisfaction function to video streaming in the form of a curve.

13 is a graph illustrating the distribution of the cumulative density function of packets with respect to the delay time of video streaming traffic when there is one video user, comparing the present invention with the prior art.

14 is a graph illustrating the distribution of the cumulative density function of packets with respect to the delay time of WAP traffic when there is only one video user, comparing the present invention with the prior art.

FIG. 15 is a graph illustrating the distribution of the cumulative density function of packets with respect to the delay time of video streaming traffic when two video users are compared with the present invention.

FIG. 16 is a graph illustrating the distribution of the cumulative density function of packets with respect to the delay time of WAP traffic when two video users are compared with the present invention.

Hereinafter, with reference to the accompanying drawings will be described in detail the operating principle of the preferred embodiment of the present invention. In the following description of the present invention, detailed descriptions of well-known functions or configurations will be omitted if it is determined that the detailed description of the present invention may unnecessarily obscure the subject matter of the present invention. Terms to be described later are terms defined in consideration of functions in the present invention, and may be changed according to intentions or customs of users or operators. Therefore, the definition should be made based on the contents throughout the specification.

The present invention to be described later is to efficiently schedule the transmission of data packets on the downlink in a wireless communication system. More specifically, the present invention relates to scheduling by the base station's scheduler and the allocation of wireless traffic channels accordingly. The present invention is performed by a scheduler of a base station constituting a wireless communication system based on a time division multiplex (TDM) scheme of a shared medium.

2 illustrates a wireless communication network model to which the present invention is applied. As shown, the entire service area is divided into a plurality of small service areas 1 called cells, and each cell is served by a base station 2 performing wireless communication.

3 is a diagram illustrating a system configuration of the wireless communication network shown in FIG. 2.

Referring to FIG. 3, a wireless communication system includes base stations (MSs) 11 and 12 and base stations communicating wirelessly with the terminals 11 and 12 and communicating with them through a wireless channel. Subsystems (BTSs) 20,30 and Base Station Controllers (BSCs) 40 connected and in communication with the base stations 20,30. The base station controller 40 is connected to a Mobile Switching Center (MSC) 50 and also to a gateway 60 (GW) 60.

The mobile switching center 50 is connected to a circuit network such as a Public Switched Telephone Network (PSTN), and the gateway 60 is an Internet / Public Switched Data Network (PSDN). It is connected to a packet switched network such as. The terminals 11 and 12 are connected to the mobile switching center 50 or the gateway 60 under the control of the base station controller 40.

The structure shown in FIG. 3 is a generalized structure of a wireless communication system, and the names of the components indicate which system (eg, IS-2000, WCDMA, UMTS, CDMA2000 1xEV-DO, GPRS, 1xEV-DV, etc.). For example, the gateway 60 is a logical name and may be called other names such as a packet data service node (PDSN), an access gateway (AG), a media gateway, and the like. As another example, the gateway 60 and the mobile switching center 50 may be integrated into the same system.

In the wireless communication system configured as described above, call setup and data transmission between the terminals 11 and 12 and the base stations 20 and 30 are performed according to the service class of the user. The service class is defined as a band allocation class for a channel bearer and a queuing class for traffic control when the terminal first establishes a call in the wireless communication system. In particular, in the present invention, the service grade is used to determine a user's satisfaction function.

In order to service a plurality of terminals in a base station, scheduling should be performed in consideration of the satisfaction of terminals having data to be transmitted every scheduling period. For this scheduling operation, the base stations 20 and 30 include buffers 21 and 31 and schedulers 21 and 31, respectively.

4 is a diagram illustrating a configuration of a base station (BTS) 20 illustrated in FIG. 3. It is assumed here that the base station is base station 20 of FIG.

Referring to FIG. 4, the base station 20 includes a main processor 210, a line interface 220, an intra-BTS switch or router 230, and channel cards 241. 243, a radio frequency (hereinafter referred to as RF) includes a transmitter / receiver 250 and a scheduler 21. The master controller 210 generally controls the base station 20. The line interface 220 is for connection with the base station controller 40. The RF transceiver 250 is for transmitting and receiving data and control signals in the form of a radio signal (RF signal) between the terminal 11. The switch 230 determines the traffic path within the base station. The scheduler 21 supports efficient management of radio resources.

5 is a diagram illustrating the configuration of channel cards 241 to 243 shown in FIG. This configuration will be assumed to be channel card 241, but the same applies to the other channel cards 242 to 243.

Referring to FIG. 5, the channel card 241 includes an input / output interface 24-1, a main processor 24-2, a memory 24-3, and at least one modulator 24. -4 and at least one demodulator 24-5. The input / output interface 24-1 is for connection with the switch 230. The modulator 24-4 modulates data and control signals to be transmitted to the terminal 11 through the transmitter 251 of the RF transceiver 250. The demodulator 24-5 demodulates data and control signals received from the terminal 11 through the receiver 252 of the RF transceiver 250. The modulators 24-4 and demodulators 24-5 are also referred to as channel elements (CE) in that they each form one forward channel or reverse channel. The memory 24-3 includes an internal buffer (buffer 22 of FIG. 3) that temporarily receives packet data to be transmitted to the terminal 11 from the base station controller 40. In addition, the memory 24-3 may store various control information.

2 to 5, in a wireless communication system using a time division scheduling scheme for data services, a delay felt by a user is scheduled to allocate a data packet to a slot according to a channel characteristic of the user. In addition to the scheduling delay time, processing delay time to wait while serving another user's data packet should be included.

In other words, the delay until the time when the entire data packet requested by the user is fully serviced can be classified as follows.

Latency = Access Latency + Media Transfer Time + Scheduling Latency + Processing Latency

In this case, the access delay is a delay time required for the wireless user to request upward and is not related to the scheduling scheme. The medium propagation delay may also be neglected as a delay time for passing through a wired or wireless medium. There are two types of delays that affect scheduling schemes: scheduling delays and processing delays. , , The latency you feel ^T is approximated as in Equation 3 below.

Where processing latency Is determined by the length of the packet the user will be served and the channel status of that user. If the channel status does not change, then the user b receives a predetermined bit rate. Is allowed and the size of the data packet requested by user j And the time in one time slot In this case, the expected processing delay time of this user is given by Equation 4 below.

Comparing this with Equation 2 mentioned above, the transmission rate given to each user In addition, the length of the packet requested by the user It can be seen that is considered.

Users who want to receive data service have different satisfaction level according to the delay time until the required packet is completely received and the transmission rate depending on the channel condition. According to Equation 4, the requested transmission rate of the user Is Is given as a function of Can be said to be a function of. I.e. user j's satisfaction function If this user's satisfaction function Is expressed by Equation 5 below.

Said satisfaction function Can be variously set according to the service level of the user. As an example, upon new subscription of a user, the system operator sets up a unique satisfaction function according to the requested service type of the user.

FIG. 6 is a diagram illustrating a utility function set according to a delay time for each user according to the present invention in a curved form. As shown, curve a is convex, curve b is linear, and curve c is It is concave.

Here, curve a is more sensitive to delay time than curve c, and curve b is more sensitive to delay time than curve c and less sensitive than curve a. In general, the satisfaction that each user feels about the delay time decreases with time, and as the delay increases, the satisfaction does not drop significantly. In the case of delay-sensitive traffic, when a certain time passes, satisfaction decreases rapidly as shown by a curve in FIG. 6. Such a curve a is applicable to, for example, a user who wants a real-time video service.

The scheduler must determine which user to assign the current time slot to every time slot. However, the wireless channel does not maintain a constant state due to fading due to path loss, adjacent obstacles, and reflection effects depending on the distance between the user and the base station. In addition, the channel state is unpredictable because the Doppler effect is caused by the mobility of the user and the path loss and fading are further affected.

Therefore processing latency Is changed according to channel conditions and user satisfaction may also vary depending on scheduling time. In addition, in the time division scheduling technique, the processing delay time of one user j Processing latency for other users k In this case, which user is assigned a slot at what time point also affects the satisfaction of all other users. Therefore, when a slot is allocated to a user at a certain scheduling time, the sum of satisfaction of all users should be scheduled to be the maximum. Therefore, when a packet is serviced to a user j at the current scheduling time based on the current channel state, a change in the satisfaction value due to this should be calculated. This may be expressed as Equation 6 below.

here Denotes a satisfaction function of user j.

If the current channel state is maintained, assigning slots consecutively to a particular user increases the sum of the user's potential satisfaction. Explaining this, there is a user i who is continuously assigned a slot and the sum of potential satisfaction is Let's assume that the maximum. If one slot is currently assigned to another user j with a processing delay of two or more slots without sequentially assigning slots, the processing delay of user i originally intended to be allocated consecutively Will be pushed out by one slot. In other words, the sum of potential satisfaction is shown in Equation 7 below.

here Is the time of one slot.

If the processing delay of user j is one slot, it is obvious that the first service to user j is greater than the original scheduling method. Violates the initial setting of max.

This will be described in more detail with reference to FIG. 7. In (a) illustrating continuous scheduling, a delay time required for user 1 to receive all of the desired data packets is Delay time required for user 2 to receive all data packets to be. On the other hand, in the case of (b) showing discontinuous scheduling, the total delay time is one slot time because User 1 has not been allocated slots continuously. Increased by Increased, but since User 2 also has not been allocated slots consecutively, Was maintained.

As described in detail above, the user satisfaction according to the delay time is a monotonically decreasing function, so assuming that the channel state is constant, consecutive slot assignments to one user are more efficient than allocating slots to multiple users. It can be seen that the satisfaction level becomes greater.

Given that N users wanting data service at a certain scheduling point are communicating with the base station, and each user has a queue, scheduling at any given point in time for the remaining packets is a priority for all users. You can also calculate whether the sum of satisfaction can be maximized. However, these calculations greatly increase the complexity of the calculation according to the number N of users connecting to the base station and data communication, and the burden of new calculations must be made at every point in time to adapt to changing channel conditions. There is this. Therefore, in the present invention, an algorithm for calculating an average of a gain of satisfaction obtained by assigning a resource to one user first and a loss of satisfaction of another user will be used.

At any scheduling time, processing delays for each total packet remaining in each user's queue , , , ..., And scheduling latency ,, , ..., If the user j is serviced first, the expected satisfaction of user j is to be. However, if the user j is serviced first, the scheduling delay time of the other user k increases, so the loss of satisfaction to be. It is unpredictable which user will be served after user j finishes the service. Therefore, the expected value of loss for all users is calculated as in Equation 8 below.

Considering the above Equation 8, the change in the satisfaction agreement of all users expected when the user j is serviced first at time t Can be written as in Equation 9 below.

So at every scheduling point If the time slot is assigned to the user having the maximum value among the first values, the sum of satisfaction of all users can be maximized.

Hereinafter, a downlink scheduling operation according to the present invention will be described with reference to FIG. 8. It should be noted that the operation illustrated in FIG. 8 is performed by the scheduler of the base station. The base station also connects communication with at least one terminal (user) and provides data services. In addition, users set different satisfaction levels according to a service level when joining a data service or initiating communication.

Referring to FIG. 8, every time a scheduling period (time slot of a transport channel) starts (S105), the scheduler of the base station determines whether there is a user who is communicating with the base station and a user wants to receive data. If no such user exists, the current slot is left idle. If only one such user exists (S115), the current slot is allocated to be available to that user (S120), i.e., the base station transmits the requested data to the one user in the current slot. On the other hand, if there are two or more users, scheduling according to steps S125 to S155 is started.

In step S125, a variable j for identifying users who are communicating with the base station and wants to receive data and a variable for identifying satisfaction level for each user Are all set to their initial value 0, and the parameter max_ Above It is set to, and the parameter max_user is set to j. In step S130, j is increased by 1, and in step S135, when allocating a current slot to user j, a sum of satisfaction of all users is added. Calculate Where the sum of all users' satisfaction Is calculated by Equation 9 mentioned above.

The calculated in the process (S140) Max_ Is compared with. if Max_ If less than or equal to, it is checked in step S150 that j reaches the number of users. If it has not been reached, the process returns to step S130 and if it reaches, the process proceeds to step S155. On the other hand in the process (S140) Max_ If greater than, max_ in the process (S145) Calculated above It is reset to and max_user is set back to j and proceeds to step S150 to check whether j has reached the number of users. If it has not been reached, the process returns to step S130 and if it reaches, the process proceeds to step S155.

If the above process is repeated until j reaches the number of users, max_user indicates the user having the maximum sum of satisfaction. Therefore, in step S155, the current slot is allocated to the user indicated by max_user.

However, if there are a large number of users who require scheduling, the average data rate is poor due to poor channel condition. For small users, the processing delay over time Becomes large. In this case, the fairness of the difference in yield between users having the same satisfaction function may be a problem. This is because the total user satisfaction is considered. Therefore, even if the total yield is slightly sacrificed, such a user can be compensated for service. Arguments considered to user j for fairness Then, Equation 6, which is an object of scheduling according to the present invention, can be rewritten as in Equation 10 below.

Where above Is a fairness factor of user j, and may be set to, for example, a predetermined value predetermined for a user who has not been allocated slots continuously for a predetermined number of slots.

Hereinafter, the simulation results of the scheduling according to the present invention described above will be described with reference to FIGS. 9 to 16.

The simulations were conducted in the cellular environment of a Code Division Multiple Access (CDMA) system using a band of 1.25 MHz and using a frequency of 2 GHz. An approximate hexagonal cellular model as shown in FIG. 2 was applied and considered up to six adjacent cells, where the cell radius is 500 m. For the channel environment, Rayleigh Fading with a path index of -4 and a speed of 3 km / h was considered. The length of one slot for transmitting data is 1.67 ms, and the rate control is performed without power control, and the base station uses 20 W transmit power. The data rates are discontinuous values of 38.4, 76.8, 102.6, 153.6, 204.8, 307.2, 614.4, 921.6, 1228.8, 1843.2, 2457.6 kbps, as used in CDMA high data rate (HDR) systems Large bit rates are available. At this time, the base station does not allocate another code for the remaining power, but allocates only one slot to one user.

Using the HTTP (Hypertext Transfer Protocol) model in the wired network as a traffic model is not suitable for the characteristics of the wireless network, so the WAP (Wireless Application Protocol) traffic model is used here. In the WAP model, when a session is initiated by a user to make a request for a service, the WAP request packet is 76 bytes in size, and the time it takes for the response packet to reach the WAP gateway for the requested data is 2.5 seconds. (Exponential distribution). Several packets arrive from the WAP gateway for one request, and each packet has a Pareto distribution as shown in Equation 11 below with an average size of 256 bytes and a maximum size of 1400 bytes.

Where k is 71.7 and α is 1.1.

In the present invention, since it is about downlink scheduling, it does not consider the time required for the terminal to request data service, and there is no request of the wireless user to continuously generate traffic, and the downlink packet is continuously generated.

On the other hand, in the simulation of delay-sensitive video streaming traffic, each video frame consists of 8 packets, and each packet has a maximum size of 125 bytes and K = 20 bytes, It has a limited Pareto distribution of = 1.2. The time interval between each packet arrival is up to 6ms, K = 2.5ms, It has a limited Pareto distribution of = 1.2 and the start time interval of each image frame is 100ms. In addition, video streaming traffic will continue to occur.

Throughput is the average packet size sent in bytes per time slot, and delay is expressed in msec. The arrival time interval of a user's packets has a Poisson distribution. By adjusting this average value, the utilization of the scheduler can be obtained. The traffic load is an increase in traffic distribution based on the load when this value is less than one.

FIG. 9 is a diagram comparing the performance of the scheduling algorithm according to the present invention with respect to each satisfaction (curve a, b, c) given when the traffic load is 1.0 in FIG. Cumulative density function (CDF) graphs for. 9 to 11, HDR (High Data Rate) represents a prior art using proportional fairness without considering delay time, concave represents curve c of FIG. 6, and convex represents curve a of FIG. 6. And linear represents the curve b in FIG. 6. The same meaning is used in FIGS. 10 to 11.

As shown, using the scheduling algorithm according to the present invention, 80% of packets are completed in only 17 slots, but in the prior art, the service is completed only at 104 slots. In addition, the distribution of the cumulative density function with respect to the delay time shown in FIG. 9 may be considered as normalized throughput. As can be seen in FIG. 9, the present invention can service most packets in a faster time than the prior art, and the present invention does not drop much more than HDR even in overall yield, and the average delay time also shows better performance.

The present invention calculates, on average, the gain for the satisfaction that a user gets when first assigning a resource to one user at every scheduling time point, and the loss for the satisfaction that the other user loses. However, as mentioned, if there are N users, the calculation of O (N!) Is calculated to determine in which order all these users will be served at all scheduling points in order to maximize the sum of the potential satisfaction of all users. need. Comparing the case of using such an algorithm (hereinafter referred to as MAXU) and the satisfaction of the present invention, a result as shown in Table 1 below can be obtained.

Traffic load Prior art The present invention MAXU 0.8 0.878 0.892 0.889 1.0 0.694 0.806 0.808 1.2 0.552 0.833 0.840 1.4 0.444 0.801 0.805 1.6 0.409 0.781 0.783

Referring to Table 1, MAXU shows almost similar results as the present invention, and thus the present invention maximizes the sum of user satisfaction most efficiently. In addition, when the traffic load is 0.8, the present invention shows a slightly higher result than the MAXU because the channel state is constantly changing, so it is not significant to calculate the MAXU at every scheduling point.

Looking at Table 1 in more detail, as the traffic load increases, the utilization of the scheduling algorithm increases, so that the average user satisfaction decreases. However, when the traffic load goes from 1.0 to 1.2, there is more traffic than the scheduler can handle, which increases satisfaction because it does not service itself for low priority packets.

10 and 11 show the overall yield and the average delay for the case where the user 1,2,4,5 is fixed in a certain position in the order of good channel condition and then moved in different directions with respect to the user 3. Here, WAP traffic occurred for 36 seconds and the traffic load was 0.9. In addition, the user satisfaction function used is , = 100, user 3 moves at a constant speed of 50 km / h, and the cell radius is 500 m.

More specifically, FIG. 10 illustrates Cumulative Density Function (CDF) and yield of delay time when User 3 moves from the cell boundary to the cell center. FIG. 11 shows that User 3 has a cell boundary at the cell center. If we move to, we show the cumulative density distribution and yield of the delay time. 10 and 11, the prior art 1 refers to a scheduler according to the prior art using a low pass filter (LPF) having a constant of 100 slot times (167 ms), and the prior art 2 is a one slot time. A scheduler according to the prior art using a low pass filter having a constant of (1.67 ms).

The scheduling algorithms according to the prior arts 1 and 2 are vulnerable to mobility because they produce exponential weighted averaging in consideration of the average channel state representing the past channel state. That is, as shown in FIG. 11, the delay time increases rapidly when moving from the center of the cell to the cell boundary, because the current channel state is drastically decreased compared to the change of the average channel state. Both prior art 1 and prior art 2 show similar results.

In case of moving to the cell boundary as shown in FIG. 11, the state of the channel is deteriorated, so that the prior art 2 which obtains the average channel state by giving a large weight to the current channel state has a slightly worse result in the average delay time than the prior art 1, On the contrary, since the channel state is improved when moving to the cell center as shown in FIG. 10, it can be seen that the prior art 2 shows a somewhat better result than the prior art 1. Nevertheless, as shown in FIG. 11, when the user moves from the cell center to the cell boundary, the prior arts 1 and 2 both exhibit lower performance than the present invention. On the contrary, when moving from the cell boundary to the cell center, as shown in FIG. 10, the result is somewhat similar to the present invention, but the average delay time still shows the present invention.

The simulation result shown in FIGS. 9 to 11 is a case where only the WAP traffic model is applied. 12 to 16, a simulation result to which the present invention is applied when both video streaming traffic and WAP traffic are serviced will be described.

12 to 16 illustrate that 100% of the simulations are repeated 100 times per minute for 20% of users for video streaming traffic and 80% for WAP traffic for users having the same channel state. One result is shown. Here, HDR refers to a prior art scheduling algorithm that uses proportional fairness without considering the delay time. In addition, Proposed1 means a case where the same satisfaction function is applied to the WAP traffic and video streaming traffic, and Proposed2 means a case where another satisfaction function considering the delay is applied to the video streaming traffic.

The shape of the satisfaction function applied when another satisfaction function is applied to video streaming is shown in FIG. 12. In FIG. 12, the convex curve a represents video streaming traffic and the concave curve b represents WAP traffic. In other words, the satisfaction function for WAP traffic is concave. So, the satisfaction function for video streaming traffic is Set to. Where β = γ = 100. In case of video streaming traffic, the size of one video frame is 100ms, so if the delay time is greater than 100ms, the satisfaction is 0.

13 to 16 show the distribution of packets with respect to delay time.

FIG. 13 shows the distribution of the cumulative density function of packets for the delay time of video streaming traffic when there is only one video traffic user. Compared with Proposed 1,2, the present invention shows much better performance even when given satisfaction function such as WAP traffic. This is because, due to the nature of video streaming traffic, the size of one packet is smaller than that of WAP traffic. In particular, since video streaming traffic is sensitive to latency, the convex satisfaction function shows better performance than the same satisfaction.

14 shows a cumulative density function of the delay time of the remaining WAP traffic users when there is only one video streaming user. As shown, in contrast to the improved performance of the video streaming traffic, Proposed 1,2 in the case of WAP traffic is inferior to the prior art. However, considering the characteristics of the two traffics, this result is more satisfactory even if the performance of the WAP traffic decreases slightly because the video streaming traffic is much more sensitive to delay.

15 and 16 show a cumulative density function distribution of delay times for each traffic when two video streaming traffic users are used. Comparing Figs. 15 and 16 with Figs. 13 and 14, the results of Inventive 2, which apply different satisfaction functions to video streaming traffic users, are better.

Meanwhile, in the detailed description of the present invention, specific embodiments have been described, but various modifications are possible without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined not only by the scope of the following claims, but also by those equivalent to the scope of the claims.

In the present invention operating as described in detail above, the effects obtained by the representative ones of the disclosed inventions will be briefly described as follows.

The present invention sets the satisfaction according to the delay time according to the traffic characteristics for each user, selects the user to maximize the sum of the satisfaction of all users in every scheduling period, and allocates slots, thereby minimizing the service delay time of the user and actually Improve satisfaction In addition, the present invention has the effect of improving the data throughput of the base station by scheduling efficiency.

Claims (6)

A downlink scheduling method of a wireless communication system supporting time division multiplexing to transmit data packets of a desired size to a plurality of users who are wirelessly connected to a base station. Setting different satisfaction levels for the plurality of users; Selecting a user who maximizes the sum of satisfaction of all users among the plurality of users by using the set satisfaction level in every scheduling period; And allocating a timeslot for data transmission to the selected user. The method as claimed in claim 1, wherein the satisfaction level is a function of a user's delay time. 3. The method as claimed in claim 2, wherein the user's delay time is a sum of a scheduling delay time waiting for a slot for data transmission and a processing delay time for processing a data packet to be allocated to the assigned slot. The method as claimed in claim 2, wherein the selecting of the user comprises selecting a user who maximizes the result of Equation 12 below. Where Is the sum of satisfaction of all users when user j is selected at scheduling time t, Is the satisfaction of user j, Is the scheduling delay waiting for the slot for data transmission of user j to be allocated, Is the processing delay time for transmitting the data packet of user j to the assigned slot, and N is the total number of users. A downlink scheduling method of a wireless communication system supporting time division multiplexing to transmit data packets of a desired size to a plurality of users who are wirelessly connected to a base station. Setting different satisfaction levels determined by user delay time for each of the plurality of users; Selecting a user who maximizes the sum of satisfaction loss caused by an increase in satisfaction gain of one user and processing delay time of all other users for each of the plurality of users by using the set satisfaction level in each scheduling period; and , And allocating a timeslot for data transmission to the selected user. The method of claim 5, wherein the selecting of the user comprises selecting a user that maximizes the result of Equation 13 below. Where Is the sum of satisfaction of all users when user j is selected at scheduling time t, Is the satisfaction of user j, Is the scheduling delay waiting for the slot for data transmission of user j to be allocated, Is the processing delay time for transmitting the data packet of user j to the assigned slot, and N is the total number of users.