WO2008105568A1

WO2008105568A1 - Scheduling method of digital broadcasting service

Info

Publication number: WO2008105568A1
Application number: PCT/KR2007/001036
Authority: WO
Inventors: Sang Hyuk Kang; Sujeong Choi
Original assignee: University Of Seoul Foundation Of Industry Academic Cooperation
Priority date: 2007-02-28
Filing date: 2007-02-28
Publication date: 2008-09-04

Abstract

Disclosed is a digital broadcasting scheduling method. By comparing a request ratio with a reciprocal of the total number of data objects based on requests received from clients and recorded in a server, classifying the data objects into hot objects and cold objects in preferential consideration of objects having an outstanding request which is waiting without being yet broadcasted, and scheduling a carousel so that the hot objects and the cold objects are arranged in the carousel according to the number of times of broadcastings of the hot objects and the priority of the cold objects, the data objects can be evenly broadcasted with high success probability and short response time.

Description

SCHEDULING METHOD OF DIGITAL BROADCASTING SERVICE

[Technical Field]

This invention relates to a digital broadcasting scheduling method, and more particularly, to a digital broadcasting scheduling method which is capable of reducing user waiting time and broadcasting data desired by a plurality of clients as much as possible by reflecting clients' requests on a broadcasting schedule in carousel-based digital data broadcasting.

[Background Art]

A one-way digital broadcasting service refers to broadcasting data, which are set by a server, from the server to clients one-sidedly according to a fixed rule, while a two-way digital broadcasting service refers to broadcasting data satisfying a plurality of clients, which are selected by a server that reflects clients' requests on the data. In the two-way digital broadcasting service, how to reflect the clients' requests and when to broadcast what data object are important factors to satisfy the plurality of clients. In other words, data object scheduling according to the clients' requests is very important. A digital broadcasting service has generally two scheduling methods. A first method is a periodic broadcasting scheduling that broadcasts set data objects periodically in a fixed order, and a second method is an on-demand broadcasting scheduling that broadcasts the most suitable data object at every point of time when a data object is broadcasted according to a client's request. The periodic broadcasting scheduling may be applied to both of the one-way digital broadcasting service and the two-way digital broadcasting service while the on-demand broadcasting scheduling may be applied to the two-way digital broadcasting service. Particularly, the on-demand broadcasting scheduling reflects waiting time of clients' requests, deadline of data objects desired by clients, the number of clients' requests, and so on.

However, the periodic broadcasting scheduling has a problem in that it does not consider clients' requests since a server broadcasts data objects repeatedly according to a schedule set by the server one-sidedly. In the mean time, if there are many clients' requests, the on-demand broadcasting scheduling has a problem in that waiting time of clients' requests becomes lengthened or no data object may be broadcasted although schedules may be changed.

[Disclosure] [Technical Problem]

To overcome the above problems, it is an object of the present invention to provide a digital broadcasting scheduling method which is capable of reducing waiting time of clients in a carousel-based on-demand broadcasting service by continuing to measure requests received from clients and frequently programming objects having relatively many requests into a broadcasting program.

It is another object of the present invention to provide a digital broadcasting scheduling method which is capable of satisfying many clients by programming objects having relatively few requests as well as objects having relatively many requests into a broadcasting program. [Technical Solution]

To achieve the above objects, according to an aspect, the present invention provides a carousel-based digital broadcasting scheduling method including the steps of: comparing a request ratio with a reciprocal of the total number of data objects based on requests received from clients and recorded in a server and classifying the data objects into hot objects and cold objects; determining the number of times of broadcastings of the hot objects and priority of the cold objects; determining the cold objects to be included in a carousel depending on the length of the hot objects according to the determined number of times of broadcastings of the hot objects and the length of the carousel; determining instance spaces in which the hot objects are arranged by the determined number of times of broadcastings of the hot objects; and arranging the hot objects and the cold objects in slots of the carousel according to the determined instance spaces and broadcasting.

Preferably, the step of classifying the data objects into the hot objects and the cold objects comprises classifying objects having an outstanding request.

Preferably, the carousel-based digital broadcasting scheduling method further includes the step of rearranging and indexing the hot objects in an descending series of the request ratio.

Preferably, the number of times of broadcastings of the hot objects is determined according to the following equation: /i = min{max{2, |//]}, *7} where, f, is the number of times of broadcastings of the hot objects, f; =riB, η is a request ratio, B is the length of a carousel, U represents the upper limit of the number of times of broadcastings of the hot objects, and [] means rounding off. Preferably, the step of determining the cold objects to be included in the carousel includes: comparing the addition of 1 to the sum of numbers of times of broadcastings of the hot objects with the length of the carousel; calculating the number of cold objects, wherein the calculated number of cold objects with adding 1 to the sum of numbers of broadcastings of the hot objects is equal to or smaller than the length of the carousal if the addition of 1 to the sum of numbers of times of broadcastings of the hot objects is equal to or smaller than the length of the carousel; and determining the cold objects according to the priority by the calculated number of cold objects.

Preferably, if the addition of 1 to the sum of numbers of times of broadcastings of the hot objects is larger than the length of the carousel, the number of cold objects arranged in the carousel is 0.

Preferably, the priority of the cold objects is determined in an descending series of the product of the request ratio and waiting time of an outstanding request.

Preferably, the instance spaces in which the hot objects are arranged are determined according to the following equation:

where, d_j is an instance space of a hot object, B is the length of a carousel, and fj is the number of times of broadcastings of hot objects.

Preferably, the carousel-based digital broadcasting scheduling method further includes the step of, if there exists an empty slot in the carousel after the hot objects and the cold objects are arranged in the slots of the carousel, randomly selecting a non- outstanding object and arranging the selected non-outstanding object in the empty slot.

According to the present invention, by comparing the request ratio with the reciprocal of the total number of data objects based on the requests received from clients and recorded in the server, classifying the data objects into hot objects and cold objects in preferential consideration of objects having an outstanding request which is waiting without being yet broadcasted, and scheduling the carousel so that the hot objects and the cold objects are arranged in the carousel according to the number of times of broadcastings of the hot objects and the priority of the cold objects, the data objects can be evenly broadcasted with high success probability and short response time.

[Advantageous Effects]

The present invention has an advantage of reduction of waiting time of the clients while satisfying the clients by selecting, programming and broadcasting desired objects with specified broadcasting periods.

In addition, the present invention has another advantage in that channels can be prevented from being wasted due to indiscriminate broadcasting of objects without distinguishing between high preference objects and low preference objects even though different clients desires different objects, and objects can be evenly scheduled such that waiting time of clients is not lengthened even when the number of requests increases greatly as in an on-demand-based broadcasting system other than the carousel-based broadcasting system, and only specific objects having high priority are not continuously and repeatedly broadcasted, and thus, opportunity to broadcast other objects is not lost.

[Description of Drawings] Fig. 1 is a view illustrating an example of a digital broadcasting system to which the concept of the present invention is applied.

Fig. 2 is a flow chart illustrating a digital broadcasting scheduling method according to an embodiment of the present invention. Figs. 3(a) to 3(d) are graphical views showing success probability for a scheduling method depending on variations of the average number of received requests per second and the total number of data objects when skew values are varied.

Figs. 4(a) to 4(d) are graphical views showing response times for a scheduling method depending on variations of the average number of received requests per second and the total number of data objects when skew values are varied.

Figs. 5(a) and 5(b) are graphical views showing response time and success probability, respectively, depending on the average number of received requests per second and the total number of data objects when an upper limit value U of the number of times of broadcastings of hot object is changed. Figs. 6(a) and 6(b) are graphical views showing response time and success probability, respectively, depending on a scheduling method when the maximum waiting time P of a client is changed.

[Mode for Invention]

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings where like reference numerals denote like elements throughout. The disclosed embodiments should not be construed to limit the scope of the present invention, but are provided as examples, and it will be appreciated by those skilled in the art that various changes may be made in the disclosed embodiments without departing from the principles and spirit of the invention.

To begin with, terms used in the following description are defined as follows. An object refers to various kinds of data files such as images, texts and so on to be broadcasted through data broadcasting, a request refers to sending information related to preference to a specific object from a client to a server, and a carousel refers to a set of objects broadcasted at specified periods. The present invention relates to updating the carousel. Accordingly, the carousel may be varied in its length, but its maximum length does not exceed B. The total number of data objects is N and the carousel length is B. The length of a data object is fixed to 1. (B and 1 are represented in the unit of time.)

Assuming that the maximum time when a client waits in a data broadcasting system is P, the client does not wait any longer after the maximum time P elapses.

Accordingly, if a data object requested by the client is broadcasted in the maximum time P, it is defined that the request gets a response successfully responded, and, otherwise, it is defined that the request fails to get a response. Accordingly, response success probability is defined as a ratio of total requests to successful requests, and average lapse time until an object is broadcasted after a request to the object arrives at a server is defined as response time. Assuming that the number of requests to specified data i of requests arriving at the server and recorded in the server is qi, a request ratio ri is defined as follows.

A request of the requests recorded in the server, which is not yet broadcasted and is waiting in the maximum time P, is defined as an outstanding request. W; is defined as waiting time of the earliest arrived outstanding request of requests to the specified object i.

A hot object is defined as an object having values larger than 1 /N (N : the total number of objects) and a cold object is defined as an object having values smaller than 1 /N (N : the total number of objects).

Fig. 1 is a view illustrating an example of a digital broadcasting system to which the concept of the present invention is applied.

As shown in Fig. 1, a digital broadcasting system to which the concept of the present invention is applied includes a back channel from clients 102 to a server 101.

The clients 102 transmit requests 103 to a mobile communication network, and the server

101 stores the requests 103 received from the clients 102 in its own queue and reflects the requests on a broadcasting program to be broadcasted to the clients 102.

Fig. 2 is a flow chart illustrating a digital broadcasting scheduling method according to an embodiment of the present invention.

As shown in Fig. 2, if it is determined that a request ratio V₁ is equal to or larger than a reciprocal (1/N) of the total number (N) of data objects and there is an outstanding request, based on the requests arrived at and recorded in the server, the objects are included in a set of hot objects, and if it is determined that the request ratio η is smaller than the reciprocal (1/N) and there is an outstanding request, the objects are included in a set of cold objects. That is, the data objects may be classified into the set of hot objects and the set of cold objects depending on the request ratio η (S21).

This method is a scheduling method for an on-demand broadcasting system using requests of the clients. Accordingly, there is a need to first consider objects with an outstanding request rather than to broadcast objects with no outstanding request at a point of time of scheduling calculation (carousel start).

Accordingly, of the objects with the outstanding request, the set of hot objects A_h and the set of cold objects A_c are defined as follows.

The number of elements in the set of hot objects A_h is defined as N_A and the number of elements in the set of cold objects A_c is defined as N_A .

After classifying the objects into the set of hot objects and the set of cold objects, the elements in both sets are rearranged in order to determine the broadcasting priority of the cold objects and the number of times of broadcastings of the hot objects (S22).

That is, the elements in the set of hot objects A_h are arranged in an descending series of r_h that is, r_x r₂,- --, r_Nh .

For the elements in the set of cold objects A_c, μ; is calculated according to the following equation, μ_t = η - W_n and the cold objects are rearranged in an descending series of μ. That is, if i>j, a relationship of μ_t ≥ //_y. is established (i,j = N_Ah + \,N_Ah + 2,- -,) .

Accordingly, objects with no outstanding requests belong to neither the set of hot objects A_h nor the set of cold objects A_c. Thereafter, the number of times of broadcasting of the hot objects is determined.

Here, magnitude of all objects is fixed to 1. Then, the number of times fi of broadcasting of a hot object i is determined as follows. /, = min{max{2, [//]}, U]

Where, fj =r;B, U represents the upper limit of the number of times of broadcastings of the hot objects, and [] means rounding off.

This guarantees the hot objects to be broadcasted in one carousel at least two times. Next, as methods of determining cold objects to be included in a carousel, there are a first method in which the carousel includes all hot objects and has margin slots and a second method in which the carousel includes only hot objects (S23).

Accordingly, in order to determine cold objects to be included in the carousel, the

addition of 1 to the sum of hot objects, 1+ ^/,- , is compared with the length B of

carousel to determine whether or not there remains empty slots for cold objects in the carousel with the hot objects arranged therein.

If the addition of 1 to the sum of hot objects is equal to or smaller than B, there may exist slots for one or more cold objects. On the contrary, if the addition of 1 to the sum of hot objects is larger than B, there exists no slot assigned for cold objects. According to the above-mentioned methods, first, if the carousel includes all the

hot objects and has margin slots, that is, 1+ ∑f_j ≤B, the maximum value M to establish a

relationship of ^ /_y. + M + 1 < B becomes the number M_c of cold objects in the carousel.

In this case, the hot objects from 1 to N_h (total number of Nh=M_h) are repeatedly included in the carousel f times and cold objects (M_c) from N_h+ 1 to N_h+1+M_c are included in the carousel.

The reason for comparing the addition of 1 to the sum of hot objects with the length B of the carousel, that is, determining whether or not a relationship of 1+ Y^f _j ≤B

7<≡4

is established, is that the carousel includes all the hot objects and has one or more margin slots.

For example, assuming that the carousel length B is 12 and the total number N of data objects is 20, if the number of times of broadcasting of hot objects is l(fi=3) and

2(f₂=2), the addition of 1 to the sum of hot objects is 6 which is smaller than the carousel length, 12. Accordingly, if cold objects are 3, 4, 5, 6, 7, 8, 9, 10, 11 and 12, cold objects to be broadcasted according to priority are determined to be 3, 4, 5, 6, 7, 8 and 9.

In this manner, since the hot objects occupy five slots in the carousel, and 5+ K 12, one or more cold objects may be included in the carousel. Accordingly, it is possible to provide 7 (12-5) cold object slots.

Next, if the carousel includes only the hot objects, that is, if a relationship of

1+ ∑f_j >B is established, the maximum value M to establish a relationship of

M

∑f_j ≤ B becomes the number M_h of hot objects in the carousel. Accordingly, this

7=1 carousel includes only the hot objects from 1 to M_h, and M_c = 0.

That is, if the relationship of 1+ ∑f_j >B is established, the carousel is filled up

7<≡Λ

only hot objects so that cold objects can not be included in the carousel.

In this manner, after all M_h+M_c objects are determined, an instance space d_j of a hot object is determined according to the following equation (S24). A I ^B

The following Table 1 shows an example of a carousel constructed according to the above first method. As shown in Table 1, if the carousel length B is 12 and the total number N of data objects is 20, hot object 1 is arranged three times at every four instance spaces and hot object 2 is arranged two times at every six instance spaces. Then, cold objects are sequentially arranged in empty slots of the carousel according to priority.

More specifically, after the hot object 1 having the largest request ratio rj is first assigned to slots 1, 5 and 9 at every four instance spaces, the hot object 2 has to be assigned to slots 1 and 7 at every six instance spaces. However, since the slot 1 has been already occupied by the hot object 1, the hot object 2 is assigned to the next empty slot 2.

Accordingly, the slots 1, 5, 9, 2 and 7 of the carousel are determined by the hot objects 1 and 2, and cold objects are arranged in the remaining slots 3, 4, 6, 8, 10, 11 and 12 according to priority. [Table 1]

In this manner, after constructing the carousel using only the hot objects and the cold objects, both of which have the outstanding requests, if there remains empty slots, objects having no outstanding request are randomly selected and included in the carousel. For example, object 13 having no outstanding request is arranged after the last slot 12 assigned to the cold object.

Broadcasting is conducted according to the above-constructed carousel scheduling (S25).

Next, success probability and response time of the multi frequency scheduling (MFS) method of the present invention will be described in comparison with a scheduling method using EDF (early deadline first) and modified R (the number of requests) x W (waiting time).

Here, the MFS, EDF and RxW use a FCC (Full Cycle Completion) update system, and RxW is applied to ICR/R (Immediate Cycle Restart with object Reordering).

The FCC update system is a system that constructs a new carousel after all the data of the carousel constructed when starting the carousel are broadcasted, if a point of time when the carousel is started is different from a point of time when the objects included in a current carousel are determined.

The ICR/R is to schedule data of an entire carousel at carousel starting time, and, if a scheduling calculated at an intermediate update time is different from a scheduling of a carousel which is being currently broadcasted, suspend the current carousel and broadcast a newly calculated carousel.

In addition, objects to be included in the new carousel are determined by computing the modified RxW and taking objects having large values. In this case, RxW is the product of the number of requests, which was early measured, and the waiting time Wi of an outstanding request for an object i, which is waited for the longest time. Figs. 3(a) to 3(d) and 4(a) to 4(d) are graphical views showing success probability and response times, respectively, for a scheduling method depending on variations of the average number of received requests per second and the total number of data objects when skew values are varied.

As shown in the figures, when the average number λ of received requests per second is varied to 1, 2, 5 and 10 and if the total number N of data objects is 100, 50 and 30, considering success probability and response time depending on variation of skew values, the MFS of the present invention generally shows success probability higher than other scheduling methods, and the EDF applied to the FCC shows performance similar to the RxW applied to the FCC.

In addition, the MFS method shows response times shorter than the FCC (EDF), the FCC (RxW) and the ICR/R methods.

This means that the scheduling method for evenly reflecting clients' requests on data objects while repeatedly broadcasting data having a high request ratio in one carousel shows better performance.

Figs. 5(a) and 5(b) are graphical views showing response time and success probability, respectively, depending on the average number of received requests per second and the total number of data objects when an upper limit value U of the number of times of hot object broadcastings is changed. The graphs shown in these figures are graphs when the maximum waiting time P of a client is 30 seconds, B=20, and a skew value = 1.0 since average waiting time of Internet users is typically 20 to 30 seconds. It can be seen from the graphs that, when the upper limit value U of the number of times of hot object broadcastings is varied from 2 to 10, the performance is varied between 2 and 5, with no variation in the remaining range. Particularly, generally when U lies between 2 and 3, it is observed that the response time and the success probability are remarkably varied.

Figs. 6(a) and 6(b) are graphical views showing response time and success probability, respectively, depending on a scheduling method when the maximum waiting time P of a client is changed. The graphs shown in these figures are graphs showing performances compared when the maximum waiting time P of a client is varied if the total number N of data objects is 50, the average number λ of received requests per second is 2, and a skew value is 1.0. It can be seen from these graphs that the performance of MFS shows smoothly varying curves when P is varied from 20 to 60. In other words, as P increases, the response time and the success probability have a tendency to increase.

As shown in the graphs of the figures, the MFS shows smooth variation while other methods show rapid variation, particularly, the FCC (EDF) is most sensitive to the variation of P.

The above-described method of the present invention may be embodied by a program which may be stored in a computer readable recording medium (for example, a CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.), details of which can be easily practiced by those skilled in the art and is therefore herein omitted.

Claims

[CLAIMS]

[Claim l ] A carousel-based digital broadcasting scheduling method comprising the steps of: comparing a request ratio with a reciprocal of the total number of data objects based on requests received from clients and recorded in a server and classifying the data objects into hot objects and cold objects; determining the number of times of broadcastings of the hot objects and priority of the cold objects; determining the cold objects to be included in a carousel depending on the length of the hot objects according to the determined number of times of broadcastings of the hot objects and the length of the carousel; determining instance spaces in which the hot objects are arranged by the determined number of times of broadcastings of the hot objects; and arranging and broadcasting the hot objects and the cold objects in slots of the carousel according to the determined instance spaces.

[Claim 2] The carousel-based digital broadcasting scheduling method according to claim 1, wherein the step of classifying the data objects into the hot objects and the cold objects comprises classifying objects having an outstanding request.

[Claim 3] The carousel-based digital broadcasting scheduling method according to claim 1, further comprising the step of rearranging and indexing the hot objects in an descending series of the request ratio.

[Claim 4] The carousel-based digital broadcasting scheduling method according to claim 1, wherein the number of times of broadcastings of the hot objects is determined according to the following equation: /_< = min{max{2, Lf/]}, ^} where, f; is the number of times of broadcastings of the hot objects, fi =r;B, η is a request ratio, B is the length of a carousel, U represents the upper limit of the number of times of broadcastings the hot objects, and [] means rounding off.

[Claim 5] The carousel-based digital broadcasting scheduling method according to claim 1, wherein the step of determining the cold objects to be included in the carousel comprises: comparing the addition of 1 to the sum of numbers of times of broadcastings of the hot objects with the length of the carousel; calculating the number of cold objects, wherein the calculated number of cold objects with adding 1 to the sum of numbers of broadcastings of the hot objects is equal to or smaller than the length of the carousal if the addition of 1 to the sum of numbers of times of broadcastings of the hot objects is equal to or smaller than the length of the carousel; and determining the cold objects according to the priority by the calculated number of cold objects.

[Claim 6] The carousel-based digital broadcasting scheduling method according to claim 5, wherein, if the addition of 1 to the sum of numbers of times of broadcastings of the hot objects is larger than the length of the carousel, the number of cold objects arranged in the carousel is 0.

[Claim 7] The carousel-based digital broadcasting scheduling method according to claim 1, wherein the priority of the cold objects is determined in an descending series of the product of the request ratio and waiting time of an outstanding request.

[Claim 8] The carousel-based digital broadcasting scheduling method according to claim 1, wherein the instance spaces in which the hot objects are arranged are determined according to the following equation:

B d^{j =} \ f -

[Claim 9] The carousel-based digital broadcasting scheduling method according to claim 1, further comprising the step of, if there exists an empty slot in the carousel after the hot objects and the cold objects are arranged in the slots of the carousel, randomly selecting a non-outstanding object and arranging the selected non-outstanding object in the empty slot.