Location:
Seminar Hall, RBCCPS
Title of the Lecture: Distributed abstractions, algorithms and platforms for deadline-driven IoT analysis
Description: The Internet of Things (IoT) is a distributed system – massive in its potential deployment and data acquisition scale, unique in the diversity of applications that it will run, and significant in it its impact on society. One of the characteristics of the IoT distributed system is the heterogeneity of the computing platforms and environments, be they embedded low-power devices, medium-scale gateway and mobile platforms, accelerated computing resources, or centralized cloud data centers. While the distributed data generation and control within IoT is well recognized, enabling seamless and coordinated distributed execution of analytics within this ecosystem is still an open problem. The state of the art prefers to have a centralized coordination on the cloud, as proposed by Amazon and Microsoft’s IoT fabrics, or at best distribute the execution across an edge device and the cloud rather than collaboratively leverage the hundreds if not more computing resources.
Addressing this wide gap requires a combination of: (1) programming models that users can use to define their data assimilation, processing and decision making tasks over realtime streams; (2) abstractions and policies to specify constraints and quality of service goals for the resources and composable applications; and (3) runtime platforms and scheduling algorithms to enact these distributed tasks across the heterogeneous computing platforms to meet the goals. These must encompass the diverse needs of IoT applications and the execution environment, such as resource limitations on constrained devices and the elasticity of on-demand resources; the ability to offer guarantees on time-bound completion of analytics; the ability to dynamically adapt rapidly to changing situations; energy awareness for low-power devices; and provenance for data assurance, quality and reusability.
We propose to address these novel and complex realtime data-processing needs of IoT applications through Big Data platforms over distributed computing resources. These problems will be motivated by and validated on emerging IoT applications seen in smart cities, such as water and power management, and mobile and edge platforms like Raspberry Pis and smart phones.
About the speaker
Prof Yogesh Simmhan is an Assistant Professor at the Department of Computational and Data Sciences at IISc. Previously, he was a Research Assistant Professor in Computer Engineering at the University of Southern California, Los Angeles and Associate Director of the USC Center for Energy Informatics.
His research explores abstractions, algorithms and applications on distributed systems. These span Cloud and Edge Computing, Distributed Graph Processing Platforms and Elastic Stream Processing to support emerging “Big Data” and Internet of Things (IoT) applications. His research advances fundamental knowledge, and offers a practitioner’s insight, on building scalable and resilient systems.
He has won the IEEE/ACM Supercomputing HPC Storage Challenge Award in 2008 and IEEE TCSC SCALE Challenge Award in 2012. He is a Senior Member of IEEE and ACM, Associate Editor of IEEE Transactions on Cloud Computing and a member of the IEEE Future Directions Initiative on Big Data.
Yogesh has a Ph.D. in Computer Science from Indiana University and was earlier a Postdoc at Microsoft Research, San Francisco. Website: http://www.rbccps.org/events/?event_id1=8
