Trend Analysis and Prediction with Historical Request Data for Multimedia-on-Demand Systems

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)62_Review of books or of software (or similar publications/items)peer-review

2 Scopus Citations
View graph of relations

Related Research Unit(s)

Detail(s)

Original languageEnglish
Pages (from-to)2001-2011
Journal / PublicationIEICE Transactions on Communications
VolumeE86-B
Issue number6
Publication statusPublished - Jun 2003

Abstract

Resource-demanding services such as Multimedia-on-Demand (MOD) become possible as Internet and broadband connections are getting more popular. However, as the sizes of multimedia files grow rapidly, storage of such large files becomes a problem. Since multimedia contents will generally become less popular with time, it is desirable to design a prediction algorithm so that the multimedia content can be unloaded from the server if it is no longer popular. This can relieve the storage problem in an MOD system, and hence spare more space for new multimedia flies. In this paper, we analyse the MOD viewing trend in order to understand the viewing behaviour of users and predict the viewing trend of a particular category of multimedia based on the knowledge obtained from its trend analysis. In trend analysis, two additive regression models, exponential-exponential-sum (EES) and exponential-power-surn (EPS), are proposed to improve the fitness of the trend. The most suitable model will then be used for trend prediction based on four proposed approaches, namely Fixed Regression Selection (FRS), Continuous Regression Updating (CRU). Historical Updating (HU) and Continuous Regression with Historical Updating (CRHU). From the numerical results, it is found that CRHU, which is constructed by considering historical trend and new incoming data of viewing requests, is in general the best method in forecasting the request trend of a particular category of multimedia clips.

Research Area(s)

  • Multimedia-on-Demand, Regression models, Trend analysis, Trend prediction