A Constraint-based Approach to Edge Resource Allocation for Complex Event Processing

With the development of Internet of Things (IoT), the demand for event-driven service invocation continues to increase, which is attracting interest in the techniques of complex event processing (CEP) for recognizing event patterns in the physical world from stream sensor data. To process data with lower delay, edge computing is widely used in IoT, but there are many challenges in implementing event recognition on such distributed computing systems. This paper studies an edge resource allocation problem for complex event processing; the result is an allocation scheme that allows all complex events contained in stream data to be detected in minimal time. As the intricate composition of event patterns and the large scale of the problem makes finding an optimal solution a hard problem, we formulate the problem as a constraint-based optimization problem (COP) using features of resources and events. We then propose a CEP-efficient heuristic algorithm (CEED) for efficient solution determination. Experiments conform that our algorithm can greatly reduce the time needed for problem solving while yielding good approximation performance relative to the baseline approach.