Chapter 13: Heterogeneous Cache Sizes and Non-Uniform Demands

Learning Objectives

• Model Zipf-distributed demand and heavy-tailed popularity
• Quantify when popularity caching beats MAN's worst-case analysis
• State the decentralized coded caching scheme (Maddah-Ali–Niesen 2015) and its order-optimality
• Analyze heterogeneous cache sizes: users with different $M_k$
• Derive the hybrid popularity+randomization placement for Zipf demand
• Connect the theoretical results to CDN popularity heuristics (LFU, LRU) and practical hit-ratio analysis