Document Type

Dissertation

Date of Award

Spring 5-31-2018

Degree Name

Doctor of Philosophy in Electrical Engineering - (Ph.D.)

Department

Electrical and Computer Engineering

First Advisor

Osvaldo Simeone

Second Advisor

Alexander Haimovich

Third Advisor

Ali Abdi

Fourth Advisor

Joerg Kliewer

Fifth Advisor

Ravi Tandon

Abstract

Multimedia content is the significant fraction of transferred data over the wireless medium in the modern cellular and wireless communication networks. To improve the quality of experience perceived by users, one promising solution is to push the most popular contents as close as to users, also known as the "edge" of network. Storing content at the edge nodes (ENs) or base stations (BSs) is called "caching". In Fog Radio Access Network (F-RAN), each EN is equipped with a cache as well as a "fronthaul" connection to the content server. Among the new design problems raised by the outlined scenarios, two key issues are addressed in this dissertation: 1) How to utilize cache and fronthaul resources while taking into account the wireless channel impairments; 2) How to incorporate the time-variability of popular set in the performance evaluation of F-RAN. These aspects are investigated by using information-theoretic models, obtaining fundamental insights that have been corroborated by various illustrative examples. To address point 1), two scenarios are investigated. First, a single-cell scenario with two transmitters is considered. A fog-aided small-cell BS as one of the transmitters and a cloud-aided macro-cell BS as the second transmitter collaborate with each other to send the requested content over a partially connected wireless channel. The intended and interference channels are modeled by erasure channels. Assuming a static set of popular contents, offline caching maps the library of files to cached contents stored at small-cell BS such that the cache capacity requirement is met. The delivery time per bit (DTB) is adopted as a measure of the coding latency, that is, the duration of the transmission block, required for reliable delivery. It is proved that optimal DTB is a linear decreasing function of cache capacity as well as inversely proportional with capacity of fronthaul link. In the second scenario, the same single-cell model is used with the only caveat that the set of popular files is time-varying. In this case, online caching maps the library of files to cached contents at small-cell BS. Thanks to availability of popular set at macro-BS, the DTB is finite and has upper and lower bounds which are functions of system resources i.e., cache and fronthaul link capacities. As for point 2), the model is comprised of an arbitrary number of ENs and users connected through an interference-limited wireless channel at high-SNR regime. All equally important ENs are benefited from cache capacity as well as fronthaul connection to the content server. The time-variability of popular set necessitates online caching to enable ENs keep track of changes in the popular set. The analysis is centered on the characterization of the long-term Normalized Delivery Time (NDT), which captures the temporal dependence of the coding latencies accrued across multiple time slots in the high-SNR regime. Online edge caching and delivery schemes based on reactive and proactive caching principles are investigated for both serial and pipelined transmission modes across fronthaul and edge segments. The outcome of analytical results provides a controversial view of contemporary research on the edge caching. It is proved that with a time-varying set of popular files, the capacity of fronthaul link between ENs and content server set a fundamental limit on the system performance. This is due to the fact that the original information source is content server and the only way to retrieve information is via fronthaul links. While edge caching can provide some gains in term of reduced latency, the gain diminishes as a result of the fact that the cached content is prone to be outdated with time-varying popularity.

Share

COinS