Cellular networks are witnessing an exponential traffic growth leading to an increase in Energy Consumption (EC), and having both environmental and economic impact. Recently, different approaches have been studied to build Green cellular networks focusing mainly on the Base Stations (BSs) as the access network represents 80% of the total wireless network consumption [1] [2]. One of the promising solutions for increasing throughput and reducing EC is the deployment of large antenna arrays, known as Large Scale Antenna Systems (LSAS) [3] [4] [5] [6], that can transmit highly focused beams. In the present work, we focus on the use of an advanced BS power model developed within the GreenTouch project [7] [8]. This model allows to quantify the power consumption of a reference scenario comprising multiple sites with standard BS antennas, and then compare it to a LSAS solution implementing multilevel beamforming. We then exploit the LSAS merits to get greener deployment with less BSs, some of which can be turned off as a function of the traffic demand. The coverage areas of each cell can be modified, by updating the codebook of beams. We finally investigate different network configurations which represent distinct trade-offs between EC and capacity. We propose a methodology to represent and design green policies for managing the network which select the desired operating points. Detailed simulation results illustrate the proposed methodology.

Virtual Sectorization (ViSn) aims at covering a confined area such as a traffic hot-spot using a narrow beam. The beam is generated by a remote antenna array located at- or close to the Base Station (BS). This paper develops the ViSn model and provides the guidelines for designing the Virtual Sector (ViS) antenna. In order to mitigate interference between the ViS and the traditional macro sector covering the rest of the area, a Dynamic Spectrum Allocation (DSA) algorithm that self-optimizes the frequency bandwidth split between the macro cell and the ViS is also proposed. The Self-Organizing Network (SON) algorithm is constructed to maximize the proportional fair utility of all the users throughputs. Numerical simulations show the interest in deploying ViSn, and the significant capacity gain brought about by the self-optimized bandwidth sharing with respect to a full reuse of the bandwidth by the ViS.

Vertical Sectorization (VS) consists in creating vertically separated sectors in the original cell using an Active Antenna Systems (AAS) supporting two distinct beams with different downtilts. The total transmit power is split between the two sectors, while the frequency bandwidth can be reused by each sector, creating additional interference between the two sectors. For low traffic demand, VS may lead to performance degradation, while for high traffic demand in both sectors, VS is likely to bring about important capacity gains. Hence intelligent activation policy of VS is needed to fully benefit from this feature. In this paper, we propose an approach taking advantage of the more focused downtilted beam. A dynamic alpha-fair bandwidth sharing is proposed for low and medium load. It is autonomously replaced by full bandwidth reuse for high load scenarios using a threshold-based controller. A flow-level dynamic simulator is used to numerically validate the proposed mechanisms.

The capacity of Multiple Input Multiple Output (MIMO) Rayleigh fading channels with full knowledge of channel state information (CSI) at both the transmitter and the receiver (CSI-TR) has been shown recently to scale at low Signal-to-Noise Ratio (SNR) essentially as SNR log(1=SNR), independently of the number of transmit and receive antennas. In this paper, we investigate the ergodic capacity of MIMO Rayleigh fading channel with estimated channel state information at the transmitter (CSI-T) and possibly imperfect channel state information at the receiver (CSI-R). Our framework can be seen as a generalization of previous works as it can capture the perfect CSI-TR as a special case when the estimation error variance goes to zero. In our work, we mainly focus on the low SNR regime and we show that the capacity scales as (1-α) SNR log(1=SNR), where α is the estimation error variance. This characterization shows the loss of performance due to error estimation over the perfect channel state information at both the transmitter and the receiver. As a by-product of our new analysis, we show that our framework can also be extended to characterize the capacity of MIMO Rician fading channels at low SNR with possibly imperfect CSI-T and CSI-R.

This paper focuses on the exposure to Radio Frequency (RF) Electromagnetic Fields (EMF) and on optimization methods to reduce it. Within the FP7 LEXNET project, an Exposure Index (EI) has been defined that aggregates the essential components that impact exposure to EMF. The EI includes, among other, downlink (DL) exposure induced by the base stations (BSs) and access points, the uplink (UL) exposure induced by the devices in communication, and the corresponding exposure time. Motivated by the EI definition, this paper develops stochastic approximation based self-optimizing algorithm that dynamically adapts the network to reduce the EI in a heterogeneous network with macro- and small cells. It is argued that the increase of the small cells' coverage can, to a certain extent, reduce the EI, but above a certain limit, will deteriorate DL QoS. A load balancing algorithm is formulated that adapts the small cell' coverage based on UL loads and a DL QoS indicator. The proof of convergence of the algorithm is provided and its performance in terms of EI reduction is illustrated through extensive numerical simulations.

Small cells have been identified as an effective solution for coping with the important traffic increase that is expected in the coming years. But this solution is accompanied by additional interference that needs to be mitigated. The enhanced Inter Cell Interference Coordination (eICIC) feature has been introduced to address the interference problem. eICIC involves two parameters which need to be optimized, namely the Cell Range Extension (CRE) of the small cells and the ABS ratio (ABSr) which defines a mute ratio for the macro cell to reduce the interference it produces. In this paper we propose self-optimizing algorithms for the eICIC. The CRE is adjusted by means of load balancing algorithm. The ABSr parameter is optimized by maximizing a proportional fair utility of user throughputs. The convergence of the algorithms is proven using stochastic approximation theorems. Numerical simulations illustrate the important performance gain brought about by the different algorithms.

The capacity of multiple-input multiple-output (MIMO) Rayleigh fading channels with full knowledge of channel state information (CSI) at both the transmitter and the receiver (CSI-TR) has been shown recently to scale at low signal-to-noise ratio (SNR) essentially as SNR log(1/SNR), independently of the number of transmit and receive antennas. In this paper, we investigate the ergodic capacity of MIMO Rayleigh fading channel with estimated channel state information at the transmitter (CSI-T) and possibly imperfect channel state information at the receiver (CSI-R). Our framework can be seen as a generalization of previous works as it can capture the perfect CSI-TR as a special case when the estimation error variance goes to zero. In this paper, we mainly focus on the low SNR regime, and we show that the capacity scales as (1 - α) SNR log(1/SNR), where α is the estimation error variance. This characterization shows the loss of performance due to error estimation over the perfect channel state information at both the transmitter and the receiver. As a by-product of our new analysis, we show that our framework can be also extended to characterize the capacity of MIMO Rician fading channels at low SNR with possibly imperfect CSI-T and CSI-R.