Distributed computation offloading framework for the tactical edge

Computational constraints greatly affect the performance of tactical computers that operate mostly at the edge of the network. Knowledge of real-time computational states allows these tactical computers to adjust the computational strategies to properly process the input tasks within the required time. In this paper we describe an intelligent adaptive computing strategy to enable resource constrained tactical computers to operate with maximum efficiency. We have used object recognition features and the optimization output coming from the adaptive computing engine to decide whether the incoming video feeds can be processed locally to compute situational awareness or they should be offloaded to other more capable computational resources within the reach.