KR20100041246A - Broadcasting item scheduling method for data broadcasting - Google Patents

Abstract

PURPOSE: A broadcast item scheduling method for data broadcasting is provided to schedule a hot and cold item by combining a pull-based scheme and a push-based mode by dividing the broadcasting item into a hot item and a cold item. CONSTITUTION: The request ratio of each broadcast item is calculated by collecting the number of requests of each broadcast items(S10). Each broadcasting item is divided into a hot item and a cold item according to the request ratio(S20). The current time is determined according to the broadcast interval of the hot item. If the time for broadcasting is expired, the hot item is broadcasted in a current time slot with the push-based scheme.

Description

Broadcasting item scheduling method for data broadcasting

The present invention relates to a broadcasting item scheduling method for data broadcasting, and more particularly, a downlink channel is a broadcasting channel such as DMB, IPTV, and the like, and a forward channel is composed of various communication networks such as a cellular network, a wireless LAN, and a WiBro. The present invention relates to a broadcast item scheduling method for scheduling a broadcast order of broadcast items in a data broadcast network receiving a request for transmission of broadcast items through a reverse channel.

Broadcasting is an extremely efficient way to use bandwidth in terms of transmitting data to many users at the same time. Digital multimedia broadcasting (DMB) and IPTV, which can broadcast various media, deliver various types of information such as audio, video, and data to many users.

The data referred to herein may be a type of independent information or may be additional information of a web page or video or audio.

Examples of additional information may include an actor introduction, a song lyrics, and the like. Data broadcast items, unlike audio and video, are relatively small and often broadcast repeatedly.

DMB and IPTV are developing with the convergence of broadcasting and communication systems. In recent converged networks, the downlink is known to the broadcast network, and the return channel is known as a cellular phone, PCS, WLAN, or recently developed Wibro based on high speed mobile internet. It is expected to have a dual channel in IEEE 802.16. This has evolved into an interactive mobile broadcast environment that sends requests for user's preferred broadcast items, resulting in asymmetric forms of send and receive links in terms of network bandwidth as well as channel bandwidth.

The performance of data broadcasting is greatly affected by the scheduling algorithm. In interactive DMB and IPTV, a request for a broadcast item is sent through a reverse channel, and the server 10 needs a mechanism for scheduling the broadcast item in consideration of statistics of user requests.

The above-described technology is described for understanding the background of the present invention, and does not mean a conventional technology well known in the art.

Conventional scheduling methods divide the push and pull regions in a fixed manner in bandwidth. However, a fixed portion of the bandwidth did not sufficiently accommodate the change in popularity of the data items, and the relative demand deadlines had to be taken into account.

Another scheduling method is based on a carousel in the form of a push-based scheme. However, each carousel is newly reconfigured to meet the needs of the user in order to take advantage of the push-based mechanism, which allows the server 10 to more popular broadcast items to minimize the overall estimated access time. We had to find the cut-off point between and less popular broadcast items.

The present invention was devised to improve the above-mentioned problem, and collects preference information for a broadcast item, analyzes the demand rate, and divides the hot item and cold item into hot items and cold items. SUMMARY OF THE INVENTION An object of the present invention is to provide a broadcasting item scheduling method for data broadcasting in which a push-based scheme and a pull-based scheme are combined.

Another object of the present invention is to improve the average response time and response success rate.

The present invention devised to achieve the above object is as follows.

In the broadcasting item scheduling method for data broadcasting of the present invention, a request is analyzed for each broadcasting item to classify each broadcasting item into a hot item and a cold item, calculating a broadcasting interval of the hot item, and And periodically broadcasting the hot item in a corresponding timeslot according to a broadcasting interval.

In the step of periodically broadcasting the hot item in the corresponding timeslot, the current time is determined in accordance with the broadcasting interval of the hot item to determine whether the time for broadcasting the hot item elapses, and when the broadcast time passes, the current in a push-based manner The method may further include broadcasting the hot item in a timeslot.

The step of dividing the hot item and the cold item comprises: collecting the number of requests for each broadcast item, calculating a request rate for each broadcast item, and for each of the broadcasts according to whether the request rate is greater than or equal to a reference value. And classifying an item into the hot item and the cold item.

The reference value is changeable. In particular, the reference value may be changed according to the popularity distribution of each broadcast item.

The broadcast interval of the hot item is shorter as the request rate of the hot item is higher.

In particular, the broadcasting interval of the hot item is expressed by Equation 5 below.

S _i is a broadcast interval of the hot item i, and r _i is a request rate of the hot item i.

The step of broadcasting the hot item in the current timeslot, if there is no hot item, characterized in that it further comprises the step of broadcasting the cold item in the current timeslot according to the on-demand method.

The cold item broadcasted in the current timeslot is selected according to the wait time of the cold item, and the cold item broadcasted in the current timeslot is selected first as the longest wait time of the cold item.

The cold item broadcasted in the current timeslot is randomly selected if the longest waiting time and the same cold item is at least two or more.

According to the present invention configured as described above, by collecting the preference information for the broadcast item and according to the demand rate by dividing the hot item and the cold item by combining the hot item and cold item in the push-based and pull-based scheme, the average Improve response time and success rate.

Hereinafter, a broadcast item scheduling method for a broadcast item according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a block diagram of a broadcast item scheduling apparatus for data broadcasting according to an embodiment of the present invention, FIG. 2 is a flowchart of a broadcast item scheduling method according to an embodiment of the present invention, and FIG. 3 is an embodiment of the present invention. 4 is a diagram illustrating an example of broadcasting a hot item, and FIG. 4 is a diagram illustrating an example of broadcasting a cold item according to an embodiment of the present invention.

The server for scheduling broadcast items for data broadcasting according to an embodiment of the present invention includes a request queue 20 as shown.

As shown in FIG. 2, the server 10 analyzes the request number (S10), analyzes the request number, divides the hot item and the cold item (S20), and calculates a broadcast interval of the hot item (S30). .

Thereafter, in broadcasting the hot item, the schedule is scheduled in a push-based manner (S40). If there is no hot item, the cold item is broadcast. At this time, the cold item is scheduled in a pool-based manner (S50).

Hereinafter, each process described above will be described in detail.

First, a process of analyzing a request number in order to classify a broadcast item into a hot item and a cold item will be described.

In general, a request of a user is sent from the user terminal 30 to the server 10 through a mobile communication channel.

The server 10 analyzes the request in real time by storing the request in the request queue 20. Here, a request stored in the request queue 20 is called an outstanding request. Then, the server 10 deletes the request stored in the request queue 20 from the request queue 20 when the corresponding broadcast item is broadcasted or the deadline passes, and the request deadline of the request is the server ( 10) refers to the maximum waiting time after sending.

Typically, when a request expires, the corresponding broadcast item is no longer received by the user even though it is broadcasted, so the request is deleted from the request queue 20, and the broadcast item is not broadcasted.

For reference, in the present embodiment, it is assumed that the maximum waiting time of the users is finite, and the maximum waiting time of the users is p [sec]. However, the technical scope of the present invention is applicable to the case where the maximum waiting time of users is not finite because the p value is very large. Therefore, if a corresponding broadcast item is not broadcast until p seconds after the request is sent, that is, until the deadline, the user stops waiting for a request for the requested broadcast item.

In this case, the server 10 calculates the maximum waiting time p in order to eliminate the requests that did not receive a response until the deadline.

In general, if a broadcast item was broadcast before the maximum waiting time (p), the request is called 'success', otherwise it is called 'failure'. In this specification, since the average 'response time' and 'response success rate' are used as indicators for performance analysis, the response time is defined as an average time until the user receives a broadcast item requested. The response success rate is a probability that a broadcast item is broadcasted before the maximum waiting time p for the request.

Meanwhile, the server 10 stores all broadcast items to be broadcast in a memory in the server 10, and the server 10 includes a memory for storing all N broadcast items.

In addition, the server 10 periodically collects statistics of user requests in order to schedule broadcast items.

That is, if the average request arrival rate per second is defined as λ, the bandwidth of the broadcast channel is assumed to be 100 Kbps, and the fixed size of each broadcast item is assumed to be 100 Kbits, each broadcast item occupies a time slot corresponding to 1 second. do.

For reference, in the present exemplary embodiment, the broadcast channel bandwidth is 100 Kbps or the size of a broadcast item is described as an example, but the technical scope of the present invention is not limited thereto, and may be applied to various broadcast channel bandwidths and broadcast item sizes. There will be.

For the item i (i = 1, 2, ..., N), the number of requests so far is defined as q _i . At this time, the index i is set as q _1? Q _2? Q ₃ ? ... q _N in order of decreasing popularity.

If the current waiting time of the longest-waiting broadcast item among the outstanding requests is defined as w _i , and the demand rate measured so far is defined as r _i , the demand rate r _i is normalized to q ₁ . It is calculated by the following equation (1).

Here, r _i is normalized by 0≤r _i ≤1.

As described above, when the demand rate r _i of the broadcast item is calculated, the hot item and the cold item are distinguished based on the demand rate r _i .

The following is to separate the hot and cold items item, for setting a reference value used to identify the popularity of TV item, the reference value, as a reference divided by the hot items and cold items two groups 0≤r _i ≤1, as described above Is defined as

Therefore, if the condition r _i ≥ θ is satisfied for a given item i, a hot item is satisfied, and a cold item is satisfied if r _i <θ.

If this is defined as the equation, it is the same as equation (2) and equation (3).

As shown in Equation 2, the hot item is

And a cold item is represented by Equation 3,

Is defined.

In this case, the hot item and the cold item may be differently classified according to the reference value θ. If the reference value θ is large, many broadcast items are classified as cold items. If the reference value θ is small, the hot item is classified as a hot item. There will be a lot of broadcast items.

Therefore, if the reference value (θ) is large, many broadcast items are classified as cold items, and thus are effective when the popularity of broadcast items is distributed at a relatively even value, and at this time, the broadcast items are mainly transmitted on-demand. Will be. If the reference value θ is small, the number of broadcast items classified as hot items increases, which may be effective when the popularity distribution is uneven. In this case, scheduling is mainly performed in a push-based manner.

Scheduling in a pull-based and push-based manner will be described later.

Meanwhile, the number of hot items is defined as K. At this time, if there is no broadcast item that satisfies r _i ≥ θ by making the reference value θ very large, the hot item number K becomes 0, and all items become cold items, resulting in a full-based mechanism.

Therefore, by adjusting the reference value [theta] appropriately, 1≤K≤N-1 and i = 1,2, ..., K hot item and i = K + 1, K + 2, ..., N With a cold item of, in this case, the pool-based and push-based mechanisms are broadcast in a mixture. Finally, if K = N, the hot item count corresponds to push-based scheduling.

When scheduling broadcast items, the server 10 gives priority to hot items over cold items. Server 10 uses a push based mechanism for hot items. Thus, hot items are broadcast periodically.

In this manner, after the broadcast item is divided into a cold item and a hot item according to the reference value θ, the server 10 calculates a broadcast interval for the hot item.

The broadcast interval for the hot item i is calculated by equation (4). [X] here means rounding.

As more hot items are requested, the server 10 broadcasts broadcast items with short broadcast intervals.

Here, Equation 4 is applied only to the hot item and leaves the cold item broadcasted in a pull-based manner to obtain better performance when the broadcast item has a relatively even popularity.

As described above, after calculating the broadcast interval for the cold item, scheduling for the hot item and the cold item is performed.

Before describing the process of scheduling hot and cold items, the summary of each notation is as follows.

Each notation is

N: total number of broadcast items

λ: average request arrival rate

P: Maximum waiting time for user

q _i : Total number of requests for broadcast item i

q _i ^* : Number of requests for outstanding item i

r _i : Request rate for broadcast item i

w _i : Longest waiting time for broadcast item i of the requests in the current request queue.

θ: Reference value for distinguishing hot item from cold item

A _h and A _c : hot and cold item sets, respectively

K: number of hot items

s _i : Broadcast interval of hot item i

k: current time slot

b _i : Time slot when the broadcast item i was last broadcast

to be.

In the present embodiment, it is assumed that one broadcast item occupies one time slot in a broadcast channel, and since the last broadcast time slot for the hot item i is b _i , the next broadcast time of the next broadcast item i is a time slot. b _i + s _i is scheduled.

Assuming that the current timeslot is k, the current timeslot

k <b _i + s _i

This means that the broadcast time for the hot item i has not been reached.

And the current timeslot

k≥b _i + s _i

In this case, the broadcast item i has already passed the time that should have been broadcast. Therefore, the server 10 should broadcast the broadcast item i as soon as possible. In each timeslot, the server 10 first finds a hot item i ^* that satisfies Equation 5 below.

If the server 10 finds a broadcast item i that satisfies a condition as in Equation 5, that is, k≥b _i + s _i , the server 10 sends the broadcast item i ^* to the current timeslot.

If there is no hot item that satisfies Equation 5, then the server 10 finds a suitable cold item for broadcasting in an on-demand time slot according to the following procedure.

The server 10 searches the current request queue 20 to find the cold item i ^* having the largest value according to Equation 6 below.

The server 10 randomly selects two or more cold items satisfying Equation (6). Server 10 then sends a cold item i ^* to the current timeslot.

3 shows an example of broadcasting a hot item, when the hot items are 1,2,3, the first broadcast item of these items has a next broadcast schedule still behind the current time k, and the second broadcast item coincides with the current time. In addition, since the broadcast item 3 is a broadcast item to be broadcast most quickly because the time for broadcasting has already passed, in this case, the broadcast item 3 is broadcast in the current timeslot k.

In addition, FIG. 4 shows an example of determining that a cold item is to be broadcast in the current timeslot since the next broadcast time is later than the current time as a result of examining the hot items. At this time, the cold item to be broadcast is determined by Equation 6, and in the example of FIG. 3, the broadcast item 4 satisfies such a condition.

5 is a view showing a simulation result according to an embodiment of the present invention.

Looking at the results of the simulation according to this embodiment as follows.

Assume that requests from users to the server 10 enter the Poisson process. Simulation parameters are in Table 1. It is assumed that the popularity of broadcast items follows a Zipf distribution, where the Zipf parameter is assumed to have a value between 0.6 and 1.4 in the normal range.

Simulation parameters Zipf skew coefficient 0.6 to 0.4

)Poisson request arrival rate (

) 2 and 5 simulation run time 1,000,000 time slots (sec) total number of items (

) 100 and 500 item size 100 Kbits client max waiting time (

) 20 to 60 [sec] and

channel bandwidth 100 Kbps (DMB channel)

In general, response time and response success rate are used as performance indicators of the scheduling method. Response time refers to the average time taken by the user to send a request and receive a broadcast item accordingly. And, the response success rate means the probability that the user's request will be broadcast within the deadline.

Considering that the maximum response time of the user is about 20-30 seconds when using the Internet, the simulation was performed assuming that the user did not wait more than 30 seconds after sending the request with p = 30 [sec].

The server 10 reflects the maximum response time of the user as a design parameter of the system. In this case, in the case of a non-broadcast request of a broadcast item, the response time corresponds to infinity, and the actual finite response time does not exceed 30 sec or more.

In the present Example, it observed in the state set to (theta) = 0.1. Since the response time is already limited to a maximum of 30 seconds, the success rate is compared for performance comparison.

Referring to FIG. 5, it can be seen that the scheduling method according to the present embodiment has a high success rate. Since the maximum response time is limited, the smaller the user request arrival rate λ, the smaller the success rate.

In this embodiment, the hot item is broadcasted according to a predetermined interval, and the broadcast interval is determined by the request value of the hot item, and the cold item is broadcasted in a time slot in which there is no hot item.

The conventional scheduling method is a carousel-based method, in which a server broadcasts according to a first-come-first-served (FCFS) method whenever a user's request is prioritized with an on demand item first. When not present, the server periodically broadcasts hot items. This considers cold items first.

However, in this embodiment, the hot items are broadcast at predetermined intervals, and the intervals are not disturbed by the cold items.

As a result, the present embodiment can obtain a small response time while maintaining a high success rate.

Although the present invention has been described with reference to the embodiments shown in the drawings, it is merely exemplary and various modifications and equivalent other embodiments are possible to those skilled in the art. Will understand. Therefore, the true technical protection scope of the present invention will be defined by the claims.

1 is a block diagram of a broadcast item scheduling apparatus for data broadcasting according to an embodiment of the present invention.

2 is a diagram illustrating a broadcast example of a hot item according to an embodiment of the present invention.

3 is a diagram illustrating a broadcast example of a cold item according to an embodiment of the present invention;

<Explanation of symbols for the main parts of the drawings>

10: server 20: request queue

30: user terminal

Claims (11)

Analyzing a request for each broadcast item to divide each broadcast item into a hot item and a cold item; Calculating a broadcast interval of the hot item; And And broadcasting the hot item periodically in a corresponding timeslot according to the broadcasting interval of the hot item. The method of claim 1, wherein periodically broadcasting the hot item to a corresponding timeslot includes: Determining whether the current time has elapsed to broadcast the hot item according to the broadcasting interval of the hot item, and if the time to broadcast has elapsed, broadcasting the hot item in a current time slot in a push-based manner. Broadcast item scheduling method for broadcasting. The method of claim 1, wherein the step of dividing the hot item into a cold item comprises: Calculating a request rate for each broadcast item by collecting the number of requests for each broadcast item; And And classifying each broadcast item into the hot item and the cold item according to whether the request rate is equal to or greater than a reference value. 4. The method of claim 3, wherein the reference value is changeable. 5. The method of claim 4, wherein the reference value is changeable according to a popularity distribution of each broadcast item. The method of claim 1, wherein a broadcast interval of the hot item is shorter as a request rate of the hot item is higher. The broadcast interval of the hot item is expressed by Equation 5 below.

S _i is a broadcast interval of the hot item i, and r _i is a request rate of the hot item i. The method of claim 1, wherein broadcasting the hot item in the current timeslot If the hot item is absent, broadcasting the cold item in the current timeslot according to an on-demand method. 10. The method of claim 8, wherein the cold item broadcasted in the current timeslot is selected according to a wait time of the cold item. The method of claim 8, wherein the cold item broadcasted in the current timeslot is selected first as the waiting time of the cold item increases. 10. The method of claim 8, wherein the cold item broadcasted in the current timeslot is randomly selected when at least two cold items with the longest waiting time are the same.

Images