Popularity-aware mobile content management: When is knowing less better?

Abstract

Changes in mobility patterns and the topology of mobile networks, the capability of mobile devices to directly serve content, and the deployment of the Internet of Things (IoT) where mobile sensors and actuators directly provide content and services, has changed the requirements and goals of traditional IP mobility management. As a result, IP mobility management and mobile content management are becoming increasingly interdependent. This paper presents and evaluates a popularity-aware lossy resolution procedure for mobility management that decides, based on its popularity and mobility, which mobile content (or mobile devices) should be tracked by the resolution system and which should be located using broadcast queries. The procedure supports the aggregation of multiple location updates in a single update message. Our analytical model jointly captures the tradeoffs that influence the performance: the tradeoff between the cost for location updates and the cost for broadcasting location queries, and the tradeoff between the reduced update cost that is achieved with aggregation and the increased delay, which results in outdated location information. Investigations show how the proposed approach’s performance, in terms of reduced signaling cost for resolution and reduced memory requirements for resolution tables, depends on the content mobility, aggregate content request rate, broadcasting cost, location update aggregation, content popularity, and correlation between content popularity and mobility. The investigations include experiments with campus YouTube traces and global YouTube video popularity.