Interference Management in Two Tier Heterogeneous Network Background Dense deployment of small cell BSs has been proposed as an effective method in future cellular systems to increase spectral efficiency and coverage. In particular, a two tier HCN architecture has been envisioned in next generation cellular systems, which consists of a single macro cell base station (MBS) and many small cell base stations (SBS). See Figure 1 below. In particular, when MBS and SBS are sharing the same spectrum (so called co channel deployment), there will be inter tier interferences. In order to mitigate inter tier interferences, state of the art approaches typically use power control scheme. However, increasing the transmission power may make the inter tier interferences decrease the total throughput of the system. Meanwhile, advanced interference management techniques such as partial interference cancellation (PIC) and interference alignment (IA) have be shown to be optimal in two user and many user interference channels in the information theoretical literatures. However, such advanced schemes are difficult to be implemented in practice since it requires channel state information among user equipment (UE) and BS. As a result, in conventional systems where all of the cells operate in downlink (DL) or uplink (UL) simultaneously, it is difficult to implement the advanced IM scheme. Figure 1: Two Tier Heterogeneous Network Proposed Interference Mitigation (IM) Approach We propose a framework for the mitigation of inter tier interference in two tier heterogeneous network under the co channel deployment, which is summarized in Figure 2 below. The key ideas are the following: (1) First, we propose reverse duplex, where the macro cell DL is combined with the small cells UL, and the macro cell UL is combined with the small cells DL. The goal is to make inter tier interference become inter BS and inter UE. In this way, inter BS interference becomes more
Figure 2: Proposed Interference Mitigation Framework structured and can be mitigated using advanced IM schemes, because base stations are typically more capable of carrying out such computationally expensive tasks. Moreover, base stations are more static and typically do not move at all, and hence the CSI measurement between MBS and SBS is more feasible and accurate. Inter UE interference, on the other hand, has very little structure and hence can be simply treated as noise. (2) Second, based on the reverse duplex architecture, we implement PIC (the Han Kobayashi scheme) in macro DL with small UL and IA in macro UL with small DL. Furthermore, we assume that the limited backhaul between MBS and SBS is able to communicate reliable transmission to assist mitigating interference. When macro is DL and small is UL, macro cell separates DL messages into common messages and private messages and small cells decode the corresponding common messages and treat the corresponding private messages as noise. The amounts of the corresponding common messages which need to be decoded at SBS depend on the strength of the interference across MBS and SBS. The limited backhaul can use codebook binning to reduce the range of the common messages for the corresponding small cell. When macro is UL and small is DL, all of the transmitters use lattice code or multiple dimensions to align the interference at MBS to mitigate a lot of strong interference from other SBSs. The limited backhaul can use bit level interference cancellation to assist increasing the system throughput. MUE MUE SU SU Macro DL Small UL: Partial Interference Cancellation (PIC) Macro UL Small DL: Interference Alignment (IA)
Results Based on the proposed framework, we characterize the information theoretical capacities to within a bounded gap for the one to many interference network (IN) and the many to one IN, corresponding to the macro DL small UL and the macro UL small DL scenarios respectively. Then, in order to verify the effectiveness of the proposed scheme, we conduct numerical simulations in a simulation environment which refer to 3GPP TS 36.211: Evolved Universal Terrestrial Radio Access (E UTRA); Physical channels and modulation. Table 1 describes the parameters of the environment. We assume there is line of sight (LOS) between BS and BS so that the corresponding fading channel is i.i.d. Rician channel. As for the small scale fading channel, we assume i.i.d. Rayleigh channels for the other link types. As mentioned before, we assume there is single antenna for each BS and UE and there is limited backhaul which can transmit C bps/hz. To be precise, we assume MBS can transmit C bps/hz to SBS via limited backhaul and vice versa. Furthermore, the perfect CSI across BSs is available for each BS. Also, our simulations assume that the traffic is full loading in total bandwidth. Table 1: BS and UE Parameters In the evaluation, we consider the following schemes: 1) TDD/FDD without IM (DL DL with UL UL hereafter): Treating all interferences as noise. 2) Reverse TDD/FDD without IM (Reverse TDD/FDD hereafter): Each cell treats all interferences as noise. The limited backhaul transmits quantized interference to eliminate interference between MBS and SBS. 3) Proposed Reverse TDD/FDD with IM (Proposed scheme hereafter): From above main results, we use PIC and limited backhaul for codebook binning in one to many INs. On the other hand, we use lattice code for IA and limited backhaul for bit level s interference elimination in many to one INs.
We first give a brief statement for the interferences in the system. In the case of DL DL with UL UL, each BS causes interference at the UEs of the other cells with DL transmission and each UE causes interference at the BSs of the other cells with UL transmission. In the case of proposed scheme or Reverse TDD/FDD, MBS causes interference at all of SBSs and each UE of the small cells cause interference at each UE of the macro cell and the other SBSs with MDL and SUL transmission. On the other hand, each SBS causes interference at MBS and each UE of the other small cells and each UE of the macro cell causes interference at each UE of the small cells. Now, we investigate the total throughput when applying different schemes. Since DL DL with UL UL and Reverse TDD/FDD do not deal with any interference, resulting in low total throughput in higher transmitting power. Proposed scheme deals with the interferences between MBS and SBS and the result is that we can get higher total throughput when transmitting power increasing. Above statements are verified by the results in Figure 3, i.e., the interferences from the BSs have significant impact for the system performance in higher transmitting power. Proposed scheme can increase the total throughput effectively when the whole system increases the transmitting power because of IM. Furthermore, if we use the limited backhauls for IM, both of proposed scheme and Reverse TDD/FDD can increase the total throughput. 120 110 100 90 Total system throughput, Backhaul: 6 bps/hz Proposed scheme Reverse-FDD/TDD Proposed scheme-with backhaul Reverse-FDD/TDD-with backhaul DL-DL with UL-UL Sum rate(bps/hz) 80 70 60 50 8% 17% 40 30 Reasonable) Tx power 20 10 20 30 40 50 60 Macro BS power (dbm) Figure 3: Sum Rate Evaluation and Comparison
About the Author MediaTek NTU Center BL 7H,Barry Lam Hall, 1 Roosevelt Road, Sec. 4 National Taiwan University, Taipei, 10617 Taiwan Tel: +886 2 3366 1836 Yuan Shuo Chang Former MSc Student, GICE, NTU I Hsiang Wang Assistant Professor, GICE, NTU