Study of Traffic Characteristics for Haptic Signals in split AI/ML control

Split control of robots in IoT systems requires a study on the feasibility of using the Edge computing framework. We consider such a scenario and propose a methodology to characterize the traffic rate-based feasibility of using Edge computing. We use the standard approach of estimating the entropy rate of a stochastic process and using its interpretation of the minimum number of bits required per sample for lossless compression. We provide a practical case study using a simulator for Allegro Hand to estimate the minimum required traffic for such a controlled robot. Our essential contribution is observing that the robot’s motion is restricted to random repetition of a few specific poses for a problem at hand. We observe, via numerical simulations, that a significant reduction in traffic could be achieved, making Edge computing more promising and feasible.