Server resource reservations for computation offloading in real-time embedded systems

Mobile devices have become very popular nowadays. They are used nearly everywhere. They run complex applications where the multimedia data are heavily processed. For example, ubiquitous applications in smart phones and different surveillance tasks on mobile robots. However, most of these applications have real-time constraints, and the resources of the mobile devices are limited. So, it is challenging to finish such complex applications on these resource-constrained devices without violating the real-time constraints. One solution is to adopt the Computation Offloading concept by moving some computation-intensive tasks to a powerful server. In this paper, we use the total bandwidth server (TBS) as resource reservations in the server side, and propose two algorithms based on the computation offloading to decide which tasks to be offloaded and how they are scheduled, such that the utilization (i.e., bandwidth) required from the server is minimized. We consider frame-based real-time tasks, in which all the tasks have the same arrival time, relative deadline and period. The first algorithm is a greedy algorithm with low complexity based on a fast heuristic. The second one is a pseudo-polynomial-time algorithm based on dynamic programming. Finally, the algorithms are evaluated with a case study for surveillance system and synthesized benchmarks.