Video analytics edge computing exploiting IoT cameras has gained high attention. Running such tasks on the network edge with low latency is very challenging since video and image processing are both bandwidth-hungry and computationally intensive.
Video analytics application’s performance is often affected by a subtle change in environment contexts, which existing solutions do not handle efficiently since IoT cameras are treated just as sensors or actuators.
In this thesis, we propose an context-aware dynamic IoT resource allocation scheme for video analytics edge computing to overcome such a limitation. For this, our scheme supports an application-aware IoT configuration by representing relationships between IoT camera resources and application service requirement and dynamically reconfigures IoT cameras in the presence of application and environment context changes.
We implement a proof-of-concept system to evaluate the proposed scheme. Experiment results show that our scheme can substantially reduce latency due to context changes.