ABSTRACT:

Context Information, or in other words information that characterizes the situation of an entity, or its surrounding is the backbone of relevant, useful, and timely smart applications. With the rapid development and widespread use of Internet of Things (IoT), including techniques to access big data (e.g., crowdsourcing), inferring and modelling situations anywhere in the world using a multitude of heterogenous data streams is extremely possible. Given IoT applications can be highly time-sensitive, there is a need to investigate techniques that facilitate access to context information in near-real time. Although data caching has been thoroughly investigated in the literature, adaptively caching contextual information (e.g., situations) possess unique challenges and opportunities for research [1]. In this seminar, we will discuss the unique research challenges and opportunities that persist when adaptively caching context information, parameterization of context caching decisions using Analytic Hierarchy Process (AHP), architectures and concepts in adaptive context caching (ACOCA), solutions that we have developed and tested using Deep Reinforcement Learning techniques to reactively cache context information such that cost-efficiency, performance-efficiency, and Quality of Context are maximized, further enhanced algorithms to proactively cache context information based on LSTM based predictive models using geo-indexing, and further research directions in this area.

 BIO:

Shakthi Weerasinghe received his Bachelor of Science (First Class Honors) degree in Information Technology from the Faculty of Information Technology in University of Moratuwa, Sri Lanka as the Valedictorian of the graduation. He is a former software engineer, former lecturer at University of Moratuwa, and currently a final year PhD candidate at the School of Information Technology in Deakin University, Australia investigating on the topic – “Adaptive Context Caching for Fast Concurrent Access in Internet of Things”. Shakthi has authored 11 peer reviewed articles in this area including those accepted in to PerCom and ACM Transactions in Internet of Things Journal, and 2 open-access articles. His research interests include Context-aware Computing, Internet of Things (IoT), Distributed Computing, & Interdisciplinary applications of Information Technology.

 Hosted by Professor Peter Reiher