MobiGATE: a mobile computing middleware for the active deployment of transport services

The use of gateway proxies is one important approach to facilitating adaptation across wireless and mobile environments. Importantly, augmented service entities deployed within the gateway proxy residing on the wired network can be composed and deployed to shield mobile clients from the effects of poor network characteristics. The usual approach to the static composition of service entities on the gateway proxy is to have these service entities interact with each other by explicitly invoking procedures on the named interface, but such a tight coupling of interfaces inhibits the flexible composition and adaptation of the service entities to the dynamic operating characteristics of wireless networks. In this paper, we present a mobile gateway for the active deployment of transport entities or, for short, MobiGATE (pronounced Mobi-Gate). MobiGATE is a mobile middleware framework that supports the robust and flexible composition of transport entities, known as streamlets. The flow of data traffic is subjected to processing by a chain of streamlets. Each streamlet encapsulates a service entity that adapts the flow of traffic across the wireless network. To facilitate the dynamic reconfiguration of the streamlets, we advocate applying the concept of coordination as the unifying approach to composing these transport service entities. Importantly, MobiGATE delineates a clear separation of interdependent parts from the service-specific computational codes of those service entities. It does this by using a separate coordination language, called MobiGATE coordination language (MCL), to describe the coordination among streamlet service entities. The complete design, implementation, and evaluation of the MobiGATE system are presented in this paper. Initial experimental results validate the flexibility of the coordination approach in promoting separation-of-concern in the reconfiguration of services, while achieving low computation and delay overheads.

1 Introduction

Ubiquitous access to data is becoming a reality due to the large-scale deployment of wireless communication services and advances in mobile computing devices. However, mobile computing environments exhibit operating conditions that differ greatly from their wired counterparts. In particular, applications must be able to tolerate the highly dynamic channel conditions that arise as users move about an environment. Moreover, computing devices often vary in terms of display characteristics, processor speed, memory size, and battery lifetimes. For mobile applications to operate effectively and optimally in such environments, the communication-related software must be able to adapt to dynamic conditions at runtime [6].