Energy-aware interference management for ultra-dense multi-tier HetNets: Architecture and technologies

Abstract

In order to improve spectral efficiency in future ultra-dense heterogeneous networks (HetNets), small cells and macro cells should be ultra-dense when deployed and dynamically overlaid. In this condition, overlaid deployment, state transition, and load migration will cause energy consumption and complex interference. Managing interference in ultra-dense multi-tier HetNets is challenging; interference can burden the network significantly when network conditions change over time. Because of the distribution of small cells and overload conditions in these ultra-dense networks, we must take into account the interference relationship among the base stations and distribution and loading conditions. An interference management scheme based on energy-aware architecture is proposed for ultra-dense multi-tier HetNets in this article. A survey is presented on energy-aware scheduling algorithms. We aim to study energy efficiency issues using graph theory and clustering. HetNets are divided into numerous interference areas, which correspond to coverage areas of each base station. Any conflict among users’ resources depends on if the interference user is located in a base station interference area, hence the use of a reinforcement-learning algorithm to optimize ongoing interference management. Given the complexity of a multi-layer network, conflict graph theory can easily identify network dynamics because it focuses on users’ received interference. Conflict graph theory can also increase resource reusability and efficiency. The proposed scheme can allocate the frequency spectrum equitably, reduce system interference, and improve throughput performance.

Introduction

In response to the demands of future mobile networks(eg. wireless communication services, business, web, and scientific applications), the major technical scenarios where 5G will focus on support are continuous wide area coverage, hot high capacity, low energy consumption, low latency, and high reliability [1]. To satisfy this demand, the ultra-dense multi-tier HetNet is regarded as a key technology to solve the problem of high data flow in mobile networks when confronted with high power and high capacity. Zhang et al. [2] proposed a “three cloud”, which is applied in the hot spots of high-capacity scenarios in ultra-dense HetNet architecture as shown in Fig. 1. The centralized control of clustering is implemented by removing part of the radio control functions of the base station to achieve interference management, radio resource coordination and mobility management. Thereby improving network capacity and providing users with the best service experience.

However, the ultra-dense deployment significantly increases interference between base stations, energy consumption, and carries high operational costs. Due to the randomness and density of small base station deployment, research on efficient interference management between base stations is necessary for energy-aware architecture. Interference management algorithms can also reduce energy consumption in ultra-dense multi-tier HetNets; thus, it has become an important topic in ultra-dense networks. As reported in [3], [4], the effective allocation of resources in the domains of time, frequency, and space can reduce inter-cell interference, improve system performance, and minimize energy consumption. When the base station receives very strong interference through ultra-dense base stations, the small base station in an ultra-dense multi-tier HetNet must be assigned to a completely orthogonal frequency band. This method can reduce interference among base stations and improve energy efficiency [5].

Unlike in the original cellular network, interference in ultra-dense multi-tier HetNets is local and irregular [6], [7], [8]. Interference not only exists in inter- and intra-cell contexts but also in inter- and intra-tier contexts. In ultra-dense multi-tier HetNets, the small base station adopt the same spectrum and RAT(radio access technology) as macro base station so as to bring challenge to the interference management and energy saving in ultra-dense multi-tier HetNets. Many scholars believe centralized control, such as cloud computing, data centers, clustering, or virtualization, works to mitigate these issues. However, to optimize energy saving, interference management in centralized control infrastructure should be energy aware. Interference in ultra-dense multi-tier HetNets environments is especially complex. The small base stations are deployed unplanned; the macro base station is used to increase power to cover the original user, resulting in greater interference [7], [9]. If the network architecture’s complexity can be reduced via graph theory and dynamic structural change according to interference conditions, the network will benefit from the aforementioned technologies while managing interference more reasonably, thus improving energy efficiency.

In this paper, an interference management scheme based on energy-aware architecture is proposed for ultra-dense multi-tier HetNets. We discuss energy efficiency issues using graph theory and clustering. We use the relationship of conflict interference to dynamic clustering that it will ensure the fairness of frequency allocation and proportional resource distribution, the conflict graph theory can easily identify network dynamics because it focuses on users’ received interference. Then we use a reinforcement-learning algorithm to optimize ongoing interference management.

In short, the main contribution of this paper can be summarized as follows.

• We formulate the problem of interference management base on conflict graph theory and power control in ultra-dense multi-tier HetNet architecture. We first use graph theory to describe the interference relationship among base stations.

• The system model and conflict graph-based dynamic interference management algorithm is presented, and we use the relationship of conflict interference to dynamic clustering to ensure the fairness of frequency allocation and proportional resource distribution of small base stations. Interference management and resource management with graph coloring based on graph theory can improve the problem of frequency spectrum wasting caused by orthogonal frequency allocation in ultra-dense networks.

• We propose a reinforcement-learning algorithm to optimize power control in cell clustering. Using the two state parameters (maximum interference set and received signal strength set) are used as inputs. Through centralized management in clustering and by determining the power level of each rule on each resource block in all base stations, a certain action power control must be chosen in the cluster.

The remainder of this paper is organized as follows. In Section II, background information and related research are discussed. In Section III, the system model and conflict graph-based dynamic interference management algorithm are presented. Simulation results and discusses are given in Section IV. Finally Section V concludes the paper.