A semi lazy bandit approach for intelligent service discovery in IoT applications
Nirandika Wanigasekara · 2015
Smart devices with incompatible protocols and APIs are increasing in everyday environments. Adaptive middleware techniques have enabled smart phones to become smart gateways between these incompatible devices by providing protocol translation services at runtime. Thus, mobile apps can easily use remote services to execute the app logic, which typically integrates services in a distributed mobile environment. However, as the number of services increase, and the user's context changes, it is challenging for mobile apps to discover relevant device services due to incompatible protocols and unknown devices. A smart agent is needed to guide service composition in mobile apps with respect to the user's changing context, so that mobile apps can seamlessly integrate services from different networks and execute complex tasks. To this end, I present a novel approach for context-aware service discovery based on semi-lazy bandit algorithms, and, present a functioning prototype as a microservice, which shows encouraging first results.