IN recent years, long-term evolution advanced (LTE-A) heterogeneous networks have been observed to offer reliable and service-differentiated communication [1]. According to reports from Cisco and Qualcomm [2], [3], global mobile data traffic will continue its fast growth, and the mobile data demand will increase by up to 1,000 times from 2015 to 2025. This growth has brought a heavy burden to existing cellular networks.In response, heterogeneous networks [4], representing the evolution of topological structure for state-of-the-art cellular networks, provide a promising paradigm that promotes system capacity and are expected to be a key fifth-generation (5G) cellular architecture.Heterogeneous networks often include a diverse set of small cells (picocells and femtocells) [5] deployed within the coverage of the traditional macrocell.
In parallel, heterogeneous networks have become a key enabler for many mobile applications such as smart meters, remote sensors, consumer devices, vending machines, and vehicular applications [6].These applications are closely associated with the emerging Internet of things (IoT), which is expected to be the next major issue in mobile systems [7].Here, the term ‘thing’ includes sensors, actuators, hardware, software, and storage spread over multiple disciplines suchas healthcare, industry, transport, and home appliances.The main objective of IoT is to maximize the communication of ‘things’ with the physical world and transfer the data collected by these ‘things’ into useful information [8].As one of most important application scenarios for future wireless networks and to accommodate a variety of communication technologies with diverse service requirements for heterogeneous users such as mobile users (MUs) and IoT nodes [9], the trend of IoT underlaying heterogeneous small cell networks is envisioned. One main challenge in both IoT and heterogeneous small cell networks is the energy limitation of end devices [10], [11].In heterogeneous networks, the traditional cell zooming and sleep mode, which are implemented by adjusting the transmission power of small cell base stations (SBSs) based on the traffic situation [12], [13], have been considered the most popular energy efficiency strategies.In [14], the authors propose a new algorithm to choose the optimal energy efficiency with a control server, which calculates the cost of connection for each mobile terminal to a BS depending on the distribution and number of mobile terminals and the cell capacity for each BS.The algorithm is used to identify the zooming level for each BS in terms of power efficiency and cell load to adjust the radius of the BS or enter the sleep mode.In [15], the authors propose a BS-centric cell zooming algorithm based on power control.The interference from neighboring cells can be mitigated by applying the power control algorithm based on the average distance between BSs.A BS-centric method is proposed wherein a BS decides to serve UEs who have a better channel condition with a higher priority so that the BSs can accommodate more UEs and hence improve the energy efficiency andthroughput. However, these results are only considered on macrocell base stations (MBSs) or SBSs to improve the energy efficiency of heterogeneous networks, and end-users are not considered.Increasing emissions are from mobile devices and radio access networks, accounting for 2 percent of carbon dioxide (CO2) emissions.Furthermore, only 15 percent of the energy is consumed for network operation, and the remaining 85 percent does not contribute to generating revenue [16].To decrease CO2 emissions, green communication has been studied to improve the energy efficiency in proportion to the escalating data rate.
In addition to cell zooming and sleep mode, energy harvesting (EH) has been proposed as a promising technology in wireless networks [17], [18], [19].In [20], the authors provide a comprehensive survey by exploiting interference as a perpetual energy source for wireless EH.The harmful interference can be converted to a helpful energy source by the receiver to support its operation and recharge the battery.Inspired by this idea, the concept of transferring information with energy, namely, simultaneous wireless information and power transfer (SWIPT), is considered a new technique to deal with the energy problem [21], [22].In [23], a power splitting optimization algorithm is proposed to simultaneously optimize both the wireless EH performance and information transmission in interference alignment networks.In [24], the authors propose an interference management approach in the OFDM system based on the strength of the interference.hen the interference is label, the interfering signal can be decoded with the desired signal. If the interference has comparable strength to the desired signal, the interference can be eliminated by the OFDM. In [25], the authors propose an iterative algorithm based on network reciprocity and minimize interference to achieve interference alignment with only the local channel knowledge. In a word, cell zooming and RF energy harvesting [26], [27], [28], [29], [30] have drawn much attention in heterogeneous cellular networks, with the purpose of reducing power consumption and enhancing system throughput performance.
Despite the aforementioned efforts, the addition of IoT to heterogeneous networks severely deteriorates the energy consumption problem since the end-users, namely, MUs and IoT nodes, have to compete for limited resources in a power-constrained hybrid structure. Massive IoT nodes will affect the overall throughput and aggravate the interference issue and energy problem in the existing cellular networks since in addition to the mobile users, the heterogeneous small cell networks need to share spectrum resources and provide wireless connectivity for IoT nodes, which speeds up the energy consumption.More specifically, in contrast to typical heterogeneous networks, we are facing several challenges: i) how to design the algorithm in order to implement cell zooming for SBSs while provisioning the QoS requirements of both MUs and IoT nodes; ii) how to formulate the utility maximization problem in such a hybrid system and determine the proper transmit power for SBSs during the process of cell zooming; and iii) how to update the clustering structure and select the proper cluster head (CH) for the conducted EH technology in response to the trade-off between the transmission rate and harvested energy.
To address the aforementioned challenges, we strive to propose an energy-efficient framework for IoT underlayingheterogeneous small cell networks. Our proposed framework tactfully exploits cell zooming and energy harvesting and is able to cope with the energy consumption problem and achieve the required energy efficiency, where the outage probability of end-users is guaranteed in the clustering-based IoT underlaying heterogeneous small cell networks.The main contributions of our proposed framework are summarized as follows. ? To achieve high energy efficiency, the so-called elastic cell-zooming algorithm (ECZA) in our framework is proposed, where PBSs and FBSs adaptively adjust the transmission power to zoom in or zoom out based on the QoS requirements of end-users, such that the seamless connectivity for both MUs and IoT nodes can be provided. Accordingly, the proposed ECZA significantly reduces the numbers of associated users and communications links such as BS-to-IoT node links, thereby reducing the whole energy consumption.
? In ECZA, deciding whether PBSs and FBSs should zoom in or out is not straightforward. Theoretically, we formulate this issue as a utility maximization problem and then show that this problem is nonconvex and NP-hard. To solve the problem, a heuristic greedy algorithm is proposed in this paper, thereby realizing the cell zooming strategy for PBSs and FBSs.
? We propose a SWIPT-CH selection algorithm for the cluster head to cluster member (CH-CM) structure in our framework for the IoT system. A power splitter is equipped at the CHs that adopt the power-splitting model to exploit intra-tier and cross-tier interference in the EH sense. By doing so, the proposed algorithm can simultaneously update the CH-CM structure, maximize the average residual energy of the IoT system and reduce the resource competition between IoT CHs and MUs.
? We conduct extensive simulations, and the results validate the effectiveness and robustness of our proposed framework, which can significantly reduce the power consumption while enhancing the energy efficiency of the IoT underlaying heterogeneous small cell networks.