LAUSR

This paper studies a scheduling problem in decentralized coded caching considering sequential and random request arrivals. The asynchrony between the requests begets the question of when to launch network-coded multicast to serve the requests when both the energy efficiency and the delay penalty are of concern (one can be traded for the other). To attain an optimal energy-delay tradeoff, we formulate this scheduling problem as an optimal stopping problem. In particular, for either the hard deadline or the linear penalty functions, we derive the structure of an optimal stopping rule and propose an iterative method to approximate the optimal rule. Simulation results demonstrate the effectiveness of the derived optimal stopping rules under various conditions, comparing them with several baseline scheduling strategies.