Gelecek Nesil Radyo Erişim Ağlarında Dinamik Kaynak Yönetimi

Abstract

In this thesis, radio resource allocation in a downlink multi-cell and multi-carrier distributed antenna system (DAS) are discussed. In these problems, performances of non-orthogonal multiple access (NOMA), multi-connectivity (MC) and joint transmission (JT) mode of coordinated multipoint (CoMP) are investigated. It is envisioned to use different radio access network (RAN) architectures for 5G and beyond. As network slicing impinges, some of the network functions will be centralized and some will be used in a distributed manner. Network functions such as RRM, NOMA, MC and CoMP can be either centralized or distributed. In this context, cRRM (centralized RRM) is examined in DD-RAN (dynamic distributed RAN) architecture in which some network functions (e.g., RRM) are centralized and some (e.g., CoMP) are distributed. Then, in the distributed RAN (D-RAN) architecture, the coordination among base stations (BS) is reduced and the performance is investigated in the case where RRM is performed in distributed (dRRM) and decentralized (decRRM) manner in each cell. However, the performance achieved through the utilization of cloud RAN (C-RAN) architecture as network functions are centralized, the diversity gain through centralized processing and the highest level of coordination is examined.

With all these architectural structures and at different levels of coordination among cells, the performance of JT-CoMP-NOMA is compared with scenarios where orthogonal multiple access (OMA) is employed for both situations whether MC is active or not. The combination of NOMA and MC has been observed to maximize system performance, but the pros-and-cons of using MC instead of using NOMA are assessed because of the increase in decoding complexity due to the utilization of NOMA.

In each cell there are geographically distributed remote radio heads (RRH) and randomly deployed low-power nodes. Joint power control and resource allocation problems are being optimized in various scenarios from simple to complex, for various cost functions and under appropriate constraints (e.g., minimum data rate, maximum power limit, fairness), resulting in how much transmit power should be used by access points for each user in each cell and resource block (RB). Energy efficiency (EE), spectral efficiency and spectral energy efficiency performances and transmission power are examined as a result of the optimization. Convex optimization is used to solve these non-convex problems, thus, the results obtained can be suboptimal.

Maximum system performance is obtained when cRRM is applied in C-RAN architecture having the highest degrees of freedom, followed by DD-RAN architecture. With the proposed dRRM algorithm, it is not achieved good performance as well as the similar results with decRRM. Here, it can be deduced that information sharing and coordination between BSs are important for the performance increase.

In addition to the literature, DAS is employed with JT-NOMA-CoMP studies. Furthermore, users are allowed to utilize all RBs, simultaneously, without matching users. Moreover, fairness constraint is added into optimization problem involving complex signal and network model. Additionally, a trade-off between maximizing sum-rate and minimizing output power offering different EE operation regions is investigated. More than that, JT-NOMA-CoMP is studied with different RAN architectures and coordination levels including MC framework.