Advanced Technologies Enabling the Efficient and Fair Coexistence Between LTE-U Systems andWiFi Networks

Deploying LTE in the unlicensed spectrum (LTE-U) is regarded as one of the most promising solutions to face significant data demand in the near future. According to regional regulations to access the unlicensed spectrums, LTE-U can be divided into two types: with listen-before-talk (LBT) and without LBT. The former type is regarded as the most promising global solution for LTE-U networks coexisting with WiFi networks and is a key feature
in the Release 13 of 3GPP, denoted as licensed-assisted access (LAA). While, the latter employs a duty cycle-based access scheme, which requires fewer modifications on the LTE side, enabling it to be deployed in the short term. The coexistence and performance optimization
between LTE-U and Wi-Fi is the major scope of this thesis.
In Chapter 3, the performance of LAA coexisting withWiFi is explored. The first major contribution is the more precise and comprehensive Markov Chain models developed to model the performance of baseline LBT and distributed coordinated function (DCF), which overcomes the limitations of current Markov Chain models. The second contribution is the contention window (CW) size based optimization scheme to maximize the LAA system
throughput while guaranteeing minimum WiFi throughput. The third contribution is the reinforcement learning-based algorithm developed to optimize the initial CWsize according to the environment, e.g., the number of cellular users, the traffic demand of WiFi users, etc.
In Chapter 4 RRM between LTE-U without the LBT scheme, i.e., duty cycle based scheme, and WiFi networks is studied. We are the first to formulate the RRM problem as
a many-to-one matching with incomplete preference lists. The major contribution is the 2- step matching-based algorithm proposed to obtain Pareto efficient energy efficiency of each CU in a computational complexity efficient manner.
In Chapter 5, the context is extended: CU can be allocated either an unlicensed band or licensed band while WUs are allocated unlicensed bands. The major contribution is the matching-based algorithm, which is extended to integration of many-to-one and one-to-one
matching to optimize the utility of each CU while guaranteeing minimum throughput of each CU and WU under various pricing strategies.