A Distributed Power-Saving Framework for LTE HetNets Exploiting Almost Blank Subframes
IEEE Transactions on Green Communications and Networking
Almost Blank Subframes (ABS) have been defined in LTE as a means to coordinate transmissions in heterogeneous networks (HetNets), composed of macro and micro eNodeBs: the macro issues ABS periods, and refrains from transmitting during ABSs, thus creating interference-free subframes for the micros. Micros report their capacity demands to the macro via the X2 interface, and the latter provisions the ABS period accordingly. Existing algorithms for ABS provisioning usually share resources
... ally among HetNet nodes in a long-term perspective (e.g., based on traffic forecast). We argue instead that this mechanism can be exploited to save power in the HetNet: in fact, during ABSs, the macro consumes less power, since it only transmits pilot signals. Dually, the micros may inhibit data transmission themselves in some subframes, and optimally decide when to do this based on knowledge of the ABS period. This allows us to define a power saving framework that works in the short term, modifying the ABS pattern at the fastest possible pace, serving the HetNet traffic at reduced power cost. Our framework is designed using only standard signaling. Simulations show that the algorithm consumes less power than its competitors, especially at low loads, and improves the UE QoS. Index Terms-LTE, Interference Control, ABS, e-ICIC where the macro does transmit data at a reduced power), and how many non-ABSs to provision in the next AP, so that all the HetNet load is carried, if at all possible, at the minimum power cost. When the network is overloaded (i.e., the capacity demands of the micros and the macro cannot be accommodated simultaneously), our framework degrades the service proportionally at all the coordinated nodes. The points of strength of our framework are the following: first, to the best of our knowledge, ours is the first framework to leverage ABS provisioning for power saving. Moreover, it is the only one that takes into account both I-ABSs and LP-ABSs simultaneously in a unified framework. We show that slight variations in the HetNet scenarios -e.g., in the position or traffic of UEs -such as those that may occur dynamically at the same timescale at which our framework operates, make different combinations of these mechanisms optimal from a power saving standpoint. Using both I-ABSs and LP-ABSs allows more energy-efficient transmissions, hence greater power savings or higher throughputs, than using either of the two mechanisms in isolation. It is well known that power saving comes at a cost in terms of increased latency: however, our framework also minimizes the latency for a given level of power saving, thus offering the best possible trade-off. Moreover, since our power-saving mechanism (i.e., keeping nodes off as much as possible) also reduces interference, it increases the transmission efficiency, which further reduces latency for a given traffic load. Second, but not less important, our framework employs only standardized signaling, i.e. nodes are assumed to know only what the standard allows them to about the status of the HetNet. This makes our framework practically implementable. Third, it does not require complex computations: both the macro and the micro nodes run only simple algorithms, that are independent of the number of users and scale well with the number of nodes. Fourth, it works under broad hypotheses, e.g., with arbitrary numbers of micros and network topologies, it accommodates a large class of power consumption models, and does not rely on a particular traffic or UE mobility model. Besides, it is orthogonal to algorithms running at both slower timescales, such as CRE bias selection for user association or network topology adaptation through selective node switching, and faster timescales, such as MAC-level scheduling. These will generate input data to our framework, which will then maximize the power saving. We evaluate our framework via simulation against a dynamic ABS provisioning scheme, showing that its power savings are remarkably higher, and that this does not come with any performance degradation: on the contrary, the cell throughput stays the same, and the user delay distribution improves considerably. The rest of the paper is organized as follows: Section 2 reports background on LTE. Section 3 reviews the related work, and we describe the system model in Section 4. Our power-saving framework is explained in Section 5, and Section 6 re-A. Virdis et al. "A distributed power-saving framework for LTE HetNets exploiting Almost Blank Subframes"