Coded caching is a valuable technique for reducing transmission rates in cache-enabled broadcast networks by leveraging coding and multicast opportunities. Practical implementation of coded caching schemes necessitates addressing challenges such as low subpacketization and demand privacy. This thesis focuses on investigating subpacketization and privacy aspects in coded caching.

High subpacketization, which involves dividing files into numerous subfiles, can hinder the suitability of coded caching schemes for typical file sizes and introduces high computational overhead. To mitigate this, we propose schemes with good memory-rate tradeoffs, ensuring linear subpacketization with the number of users. Using the framework of placement delivery arrays (PDAs), we present various constructions for coded caching schemes. Our graphical construction of 2-PDAs is based on graph factorization, while adaptations of partial Latin square constructions yield square PDAs.

We introduce lifting constructions, where smaller PDAs are combined to create larger PDAs. Blackburn compatibility, a new notion for PDAs, facilitates these lifting constructions. We present diverse families of Blackburn-compatible PDAs, including algebraic and randomized constructions. Extensive evaluations demonstrate the superior performance of our schemes, simultaneously achieving low subpacketization. Furthermore, we demonstrate the versatility of lifting constructions by casting existing coded caching schemes within this framework.

In addition to subpacketization, preserving the privacy of user demands is critical in coded caching. We examine demand privacy in a two-user two-file scenario, proposing an explicit scheme and revealing that privacy incurs rate and subpacketization penalties for linear coded caching schemes. Our findings also demonstrate the advantages of encoding cache contents in achieving demand privacy.