Graph-Based Joint User-Centric Overlapped Clustering and Resource Allocation in Ultradense Networks

Abstract

Abstract-With the increase of access point (AP) density and the exponential growth of mobile devices supported by ultra dense networks (UDNs), user-centric overlapped clustering relying on AP cooperation is becoming a promising design principle for guaranteeing the quality of service (QoS) required by each UE. This ambitious goal can be achieved by the joint design of user-centric overlapped clustering and resource allocation. In this context, both the traffic-load balancing and the limited availability of orthogonal resource blocks (RBs) have to be carefully considered in UDNs. To tackle these challenges, we formulate a joint user-centric overlapped clustering and resource allocation problem with the goal of maximizing the system's spectral efficiency (SE). To efficiently solve this problem, it is decoupled into two subproblems, which can be solved independently with the aid of our graph-theoretical framework. We first develop a novel distributed three-stage user-centric clustering solution, which is aware of the APs' traffic-loads. Then, based on the overlapped clusters constructed, we propose a novel two-stage graph-based user-centric resource allocation scheme for mitigating the resultant inter-cluster interference, despite the limited availability of orthogonal RBs. Our numerical results confirm the benefits of the proposed framework and show that it outperforms the benchmark solutions in terms of both its per area aggregated user rate (PAAR) and user rate.

A Model-Driven Deep Reinforcement Learning Heuristic Algorithm for Resource Allocation in Ultra-Dense Cellular Networks

“…Upon substituting (4), (16), (17), (18) and (19) into (15), we can solve (15) and get the unconstrained minimum point, given by…”

“…Nevertheless, they are hardly able to solve the multi-objective optimization problems. The existing literature only considers the partial optimization objectives of UDNs, which cannot improve network performance well [14,15]. Especially, with the rise of the environmentfriendly UDNs and user experience, the tradeoff among SE, EE and fairness has become one of the biggest challenges in the field of resource allocation for future UDNs [16,17].…”

User-Centric Clustering for Designing Ultradense Networks: Architecture, Objective Functions, and Design Guidelines

“…Given the challenge of solving the many-to-many matching, we propose to first solve a many-to-one matching problem in the first stage, and then complete the many-to-many matching process in the ensuing stages. For more technical details on the procedures please refer to [8].…”

Modularity-Based User-Centric Clustering and Resource Allocation for Ultra Dense Networks

“…1. It is noteworthy that each UE is involved in one and only one MUC cluster, whilst each AP may simultaneously belong to multiple MUC clusters as a result of overlapped UC clusters [9] [17]. Hence, the interference will be stronger within the MUC clusters and weaker amongst MUC clusters.…”

“…For example, the authors of [7] and [8] focused their attention on the UC clustering design and resource allocation. However, given the large number of UEs in UDNs, the limited number of orthogonal resource blocks (RBs) requires sophisticated resource allocation algorithms to be carried out across all access points (APs) based on the full channel state information (CSI) of all UEs [9]. This approach is impractical and hence efficient counter-measures have to be developed to improve the exploitation of orthogonal RBs.…”