Client-side optimizations for mobile publish subscribe systems

Abstract

Smartphones can provide a great deal of information about their users using numerous built-in sensors, such as location, local weather conditions, or current method of travel. This information has the potential to be relevant to other users, allowing them to be alerted about pertinent events in real-time, such as knowing where their family members are. This thesis proposes a design for a publish-subscribe system aimed to support sharing sensor data between mobile clients, where users can specify complex subscriptions over sensor data generated by other phones and receive alerts when a relevant event occurs. We present a number of client-side optimizations for collecting sensor data on the mobile device and communicating that data with the publish subscribe server efficiently, in order to reduce battery usage and data transmission. We then perform an evaluation of these optimizations, discuss design tradeoffs, and show which scenarios the optimizations perform best in.