An energy efficient framework for user association and power allocation in HetNets with interference and rate-loss constraints

Abstract

Dense deployment of femtocells has proved to be an effective solution to handle increasing demands of indoor mobile data. A femtocell not only helps reducing operational and capital expenditure but also improves the energy efficiency of the network. Femtocells are able to increase spectrum efficiency by manyfold by reusing the available spectrum for indoor users. However, it has been seen that traditional cell selection schemes limit the user count under femtocell. Additionally, dense deployment of femtocells comes with the cost of increased interference to the neighbouring femtocell and macrocell users. In this paper, we first analyse various cell selection schemes to improve user association and resource utilization in femtocells. Then, we focus on improving energy efficiency and throughput of femtocell based cellular networks. For this, we formulate an optimization problem that efficiently reuses macrocell spectrum in femtocells with power control while satisfying macrocell users’ interference and rate-loss constraints.

Introduction

Advancements in mobile communications have reached to an extent where users can expect ubiquitous connectivity on their mobile devices. This has led to a sudden increase in data demands from mobile users. Interestingly, it has been observed that a major fraction of data demands (nearly 80%) originates from indoor nomadic users [1]. In the last couple of years, we have seen an exponential increase in mobile data demands, especially due to the development of mobile platforms such as Android and Windows, and applications such as Gmail and Facebook which always remain connected to the Internet. With the availability of cheap, low-cost smartphones and tablets, these demands will continue to increase in coming future. According to the forecasts made in [2], an increase in mobile data traffic by 20 times is expected by the end of 2018.

To handle ever-increasing mobile data demands, it is imperative to significantly improve the capacity of wireless networks. Various solutions have been proposed over the years which lead to doubling up the wireless capacity every 30 months over the last 104 years. This results in an approximately millionfold increase in capacity of wireless networks since 1957 [3]. Breaking down these gains, the use of next-generation cellular technologies (LTE-A, WiMax) combined with better modulation and coding techniques resulted in a 5 × improvement in wireless capacity. However, these gains are limited by the received signal strength at mobile users and hence not always achievable. Another solution is the use of wider spectrum which has resulted in an approximate 25 × improvement in wireless capacity. Deployment of small cell base stations is the simplest and the most feasible solution to improve the capacity of wireless networks. It has resulted in a 1600 × gain in wireless capacity through frequency reuse with power control. However, additional base station deployment burdens operators with extra capital expenditure. Additionally, these base stations consume a significant amount of energy for their operation leading to a higher operational expenditure [4].

To handle indoor cellular data demands without significantly increasing operators’ expenditure, use of femtocells has been suggested [3]. Femtocell or Femto Access Point (FAP) is a small, low cost, low power cellular base station deployed inside users’ homes to provide better cellular coverage. The inherent low transmission power capabilities of femtocells allow efficient reuse of available spectrum without significantly increasing interference to nearby users. Additionally, indoor users get benefited by stronger signal quality, higher bandwidth, and better battery life because of the reduced uplink transmission power [5].

Since placement locations of femtocells are random, traditional network planning techniques fail to circumvent the interference introduced by them to primary macrocell and neighbouring femtocell users. The best way to eliminate interference in this scenario is to use orthogonal subchannels among macrocell and femtocell users. However, this diminishes the available spectrum to users in both tiers. Another approach is to allow femtocells to control their transmit power to minimize interference. However, lowering transmit power affects femtocell coverage and limits user association in them. This further reduces the resource utilization of already under-utilised femtocells. In current cellular network deployments, the utilization of resources (such as spectrum and transmit power) in macrocells is very high, specially during peak hours when unavailability of resources even leads to call blocking [6]. On the other hand, femtocell utilization is found to be significantly low (about 30%) owing to their low transmit power and coverage radius [7], [8]. Hence, to reap the gains of femtocell deployments, offloading of users from macrocell to femtocells is necessary. User offloading schemes help to free up expensive macrocell resources allocated to the offloaded users earlier, thereby increasing spectrum efficiency as well as energy efficiency.

In order to improve user association in femtocell, various solutions have been proposed in the literature. Recently, the concept of cell biasing for femtocells has been introduced in [5], [9], [10]. Cell biasing attempts to offload users from macrocells to smaller cells by modifying cell selection/handover criteria. The authors of [11] derived a closed form expression to calculate range expansion bias for both uplink and downlink for picocells while mitigating inter-cell interference. Performance analysis of Heterogeneous Networks (HetNets) for multiple small cell densities, bias values, and resource partitioning strategies have been discussed extensively in [12]. User association based on expected bitrate is recently suggested in [13]. This scheme shows quite an improvement in system throughput compared to reference signal and cell biasing based association. Our previous work in [14] suggested an enhancement to this expected bitrate cell selection scheme, further improving system throughput and energy efficiency. For better load balancing, various techniques have been proposed in the literature. Use of transmission power control is the preferred technique for range expansion and load balancing [15]. Authors in [16] proposed a heuristic based approach for inter-femtocell coordination while maintaining fairness. In [17] and [18], authors demonstrate that fractional transmission power control provides improved performance to cell edge users. Work in [19] adjusts femtocell transmission power by a decentralized algorithm so as to balance user load among co-located femtocells.

In this paper, for the first time a framework for joint user association and power allocation for femtocell based cellular network is proposed. The motivation behind studying both of these problems together comes from the fact that the problem of resource allocation in HetNets is immensely coupled with the problem of cell selection. A cell selection scheme may result in an unbalanced user association in HetNets. Additionally, this may result in severe co-channel interference if wireless resources such as bandwidth and transmit power are not carefully partitioned among users of different base stations. Hence, for optimal network performance, a centralised user association and power allocation framework is necessary which can jointly assign users to base stations and control interference on them using transmission power control [20]. The contributions of our work are twofold: (i) we propose the use of an energy efficient, load-conscious cell selection scheme for user offloading to femtocells and (ii) to further improve spectrum efficiency, we suggest a power control based subchannel reuse scheme for femtocell downlink. First, we examine the offloading benefits of different cell selection schemes for femtocell network and focus on evaluating network performance in terms of throughput and energy efficiency. Four existing cell selection schemes are used to offload users from macrocell to femtocells. To protect channel quality of these newly offloaded femtocell users, we suggest exclusive use of a subset of subchannels for them. Our aim is to improve energy efficiency of the network, which is achieved by reducing energy consumption as well as maximising system throughput. First, to reduce energy consumption, we offload users from high power macrocell to low power femtocells using our enhanced expected bitrate cell selection scheme. This scheme associates each mobile user to a base station which provides the maximum downlink bitrate considering the user load at base stations.

Second, efficient reuse of macrocell spectrum is proposed for femtocell downlink to maximize the overall system throughput. This, however, incurs additional interference to co-channel macrocell users. To handle this increased interference, we suggest Hybrid Constraint based Power Control (HCPC) technique. Since received signal and interference severely affect the channel quality of mobile users, we based our technique on a hybrid constraint which protects macro users based on the amount of interference it receives. Here, we divide macrocell users into two partitions; the ones those are protected by interference constraint and the rest by rate-loss constraint. To satisfy these constraints, power control over these reused macrocell subchannels is done while maximizing femto users’ throughput.

The rest of the paper is organized as follows. Section 2 presents the system model for two-tier HetNet along with user partitioning, spectrum allocation technique, energy consumption analysis, and channel model. Section 3 discusses the hybrid constraint applicable on macrocell users. We formulate our problem in Section 4 along with the interference and rate-loss constraints. Performance analysis of different cell selection schemes for heterogeneous cellular network is done in Section 5. Section 6 discusses the optimization problem formulation and the relationship between interference and rate-loss constraint. Additionally, to improve femtocell throughput, we propose a technique to effectively reuse macrocell subchannels for femtocell users. Section 7 presents the simulation scenario and obtained results. The work is concluded in Section 8, with directions for future research.