Home
Research
Publications
Teaching Experience
Professional Activities
Ph.D. Synopsis Report
M.Tech Thesis
CV
References
Contact

Research

Dept. of Computer Science of Engineering, IIT Guwahati (Jul. 2013 -Till Date)

Doctor of Philosophy (Thesis submitted on 9th August 2021).
Thesis Title: Task and Message Co-scheduling Strategies in Real-time Cyber-Physical Systems.
Supervisors: Dr. Arnab Sarkar & Dr. Chandan Karfa (Administrative Supervisor).

The principal aim of my Ph.D. dissertation has been to investigate a few important theoretical and practical aspects of task-message co-scheduling strategies in safety-critical Cyber-Physical Systems (CPSs), keeping in view the challenges/hurdles related to timing requirements, resource constraints, energy minimization, etc. The dissertation presents a few novel real-time optimal/heuristic offline task-message co-scheduling strategies for safety-critical CPSs consisting of various types of task and execution platform scenarios. In particular, the major contributions of my work include:

  • Development of co-scheduling strategies for a set of independent periodic tasks executing on a bus-based homogeneous multiprocessor system, with the objective of maximizing system level Quality of Service (QoS).

  • Design and implementation of QoS adaptive scheduling mechanisms for real-time systems modeled as Precedence-constrained Task Graphs (PTGs), on fully-connected heterogeneous multiprocessor systems.

  • Development of optimal and heuristic co-scheduling strategies for PTGs executing on a shared-bus based heterogeneous distributed platform.

  • Design of an energy-aware processor-bus co-scheduling strategy for multiple independent PTGs executing on a bus based heterogeneous platform.

Future Research Work

The static resource allocation strategies developed as part of the Ph.D. work are based on wort-case resource usage estimates of applications. These schemes may be prone to significant performance degradations when actual resource usage of the applications are significantly less than their worst-case estimates.

Research in the last few years has revealed that the mapping and scheduling mechanisms in these scenarios may need to be both static and dynamic. The static part first provides a cost-optimal constraint-driven scheduling, allocation and assignment of various functional components of all the available resources; this step should not be intended to generate a single solution, but to generate an execution plan consisting of a set of optional solutions which the dynamic part can use to take decisions according to different run-time conditions. The dynamic part should be fast and must efficiently do a combined architecture load and power-aware run-time scheduling according to the execution plan provided by the static part, such that real-time constraints are met. However, determination of the exact offline-online strategies to be employed for specific system scenarios at hand is non-trivial and demands considerable research.