Memory-Rate Tradeoff for Caching With Uncoded Placement Under Nonuniform Random Demands

For a caching system with multiple users, we aim to characterize the memory-rate tradeoff for uncoded cache placement, under nonuniform file popularity. Focusing on modified coded scheme (MCCS) recently proposed by Yu, etal., formulate placement optimization problem MCCS minimize average delivery rate popularity, restricting class of popularity-first placements. We then present two information-theoretic lower bounds one general placements and other restricted By comparing optimized bounds, prove that attains bound two-user case, providing exact tradeoff. Furthermore, it popularity-first-based case K users distinct requests. In these cases, our results also reveal is optimal MCCS, zero-padding used in incurs no loss optimality. redundant requests, analysis shows there may exist gap between due zero-padding. next fully which shown possess simple file-grouping structure can be computed via an efficient algorithm using closed-form expressions. Finally, extend study accommodate both popularity sizes, where show Numerical that, settings, only exists limited cases very small.