Dissertation Defence: Performance Analysis of Large Scale Networks in the Finite Blocklength Regime

Nourhan Soliman, supervised by Anas Chaaban, will defend their dissertation titled “Performance Analysis of Large Scale Networks in the Finite Blocklength Regime” in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Electrical Engineering.

An abstract for Nourhan Soliman’s dissertation is included below.

Examinations are open to all members of the campus community as well as the general public. This examination will be offered in a hybrid format. Registration is not required for in-person defences. Please email achaab01@mail.ubc.ca to receive the Zoom link for this defence.

---

Abstract

The emergence of delay-constrained applications impose stringent latency and reliability requirements in 5G and beyond 5G networks, necessitating a shift in the performance analysis of networks from the infinite blocklength assumption to finite blocklength. Since the blocklength in the finite blocklength regime (FBR) imposes a trade-off between delay, reliability, and information coding rate, the performance analysis under infinitely long codes would be less applicable in practical scenarios.

This dissertation presents a novel mathematical framework to characterize the performance of large-scale networks employing short codewords, conducts comparisons with the performance in the infinite blocklength regime, and investigates the effect of network and transmission parameters (e.g. base-station density, power allocation, power control parameters) on the performance. In the analysis, we consider orthogonal multiple access (OMA) and non-orthogonal multiple access (NOMA) networks. For the large-scale OMA uplink and downlink (DL) networks, theoretical achievable rates, outage probability, reliability, and the coding rate meta distribution are derived using finite blocklength coding theory and stochastic geometry. Moreover, achievable rates under practical modulation schemes and multilevel polar coded modulation (MLPCM) are investigated. Numerical results provide theoretical performance benchmarks and highlight the potential of MLPCM in achieving close to optimal FBR performance. For large-scale DL NOMA networks with a two-user cluster, the work proposes inner and outer bounds on the achievable rate region in the FBE, characterizes the min-max rate, and investigates the impact of parameters such as code-length and network density on their performance. The MLPCM is also investigated and the results validated its compliance for future technologies. Furthermore, we compared the performance of large-scale OMA and NOMA networks in the FBR, and results show the effect of network parameters, blocklength, and frame error rate on performance metrics.

Additionally, the potential of Convolutional Neural Network autoencoders (CNN-AE) in approaching the theoretical maximum achievable rate over a Gaussian channel in the FBR is investigated. The CNN-AE is compared to the theoretical maximum achievable rate and the achievable rates of other schemes from the literature, outperforming these benchmark schemes and approaching the theoretical maximum rate. This demonstrates the capability of CNN-AEs in learning good codes for delay-constrained applications.