Multi-Antenna Coded Caching for Asymmetric Content Delivery

A notable increase in internet traffic is attributed to a surge of advanced services such as holographic teleportation, brain-computer interaction, digital twins, and extended reality (XR), especially in sectors such as entertainment, education, healthcare, and culture. To meet increased demands for high data rates and minimal latency, the integration of caching with edge computing has been explored, placing computing resources closer to users and pre-storing frequently used content. However, data-intensive XR applications present challenges due to storage and bandwidth limitations for caching resources.

In this thesis, a comprehensive cache utilization framework is introduced, integrating the advantages of location-aware coded caching, multi-antenna content delivery, and device-to-device communication (D2D). The challenges posed by data-intensive applications like wireless XR environments are addressed, where users consume multimedia content while actively engaging with their virtual surroundings. Particularly, a unified location-aware multi-antenna coded caching and content delivery scheme is proposed, comprising several key elements. Initially, a novel memory allocation process optimizes the cache placement by considering users' connectivity conditions, allocating larger cache portions to the contents requested in wireless connectivity bottlenecks. The asymmetric cache placement is then utilized to design efficient codewords that enable multi-rate transmissions using multicast and/or unicast beamformers during the delivery phase. The developed algorithms strike a balance between global and local cache gains, adapting to the dynamics of volatile wireless connectivity. As a result, users in areas with poor connectivity can be allocated smaller amounts of wireless resources during the delivery phase, while users in areas with better connectivity are served at higher rates.

To benefit from closely located users having non-overlapping cache content, a D2D integrated CC scheme is also investigated. Through a well-designed two-phase transmission strategy, local D2D content exchange precedes downlink (DL) transmission, yielding two key benefits. Firstly, localized data exchange between nearby users elevates the average delivery rate, alleviating the DL transmission phase's load. Secondly, DL beamforming complexity is significantly reduced, leading to a more efficient use of computational resources. In the D2D phase, a novel low-complexity algorithm selects optimal user subsets for content exchange, balancing performance and computational efficiency. The proposed method is evaluated via simulation and mathematical analysis.