Exploring blind online scheduling for mobile cloud multimedia services

Abstract

Mobile cloud is a new emerging technology which can be used to enable users to enjoy abundant multimedia applications in a pervasive computing environment. Therefore, the scheduling of massive multimedia flows with heterogeneous QoS guarantees becomes an important issue for the mobile cloud. Generally, the predominant popular cloud-based scheduling algorithms assume that the request rate and service time, are available for the system operator. However, this assumption can hardly be maintained in many practical scenarios, especially for the largescale mobile cloud. In this article, we consider the scheduling problem for a practical mobile cloud in which the above parameters are unavailable and unknown. Taking into account the performance of the users and the impartial free time among the servers, the highlight of this article lies in proposing a blind online scheduling algorithm (BOSA). Specifically, we assign available multimedia servers based on the last timeslot information of the users' requests, and route all the multimedia flows according to the first-come- first-served rule. Moreover, we design detailed steps to apply the BOSA to a content recommendation system, and show that the proposed BOSA can achieve asymptotic optimality.