Interference Cancellation and Adaptive Demodulation Mapping Schemes for Device-to-Device Multicast Uplink Underlaying Cellular Networks

Abstract

Device-to-device multicast (D2MD) communications enable dissemination of messages directly from one user equipment (UE) to a group of D2D UEs (DUEs) located in close proximity, instead of routing via the base station. The challenges of D2MD arise from the interference to and from in-band cellular UEs (CUEs) due to frequency reuse, and the low data transmission rate constrained by the worst channel quality. To improve the overall system capacity, a novel interference coordination scheme and an adaptive demodulation mapping scheme (ADMS) based on rateless encoding technique is proposed. Firstly, a new interference limited area control scheme that reduces the interference from CUEs to each DUE is proposed. In the proposed scheme, the DUEs share same resource with the CUEs out of the interference limited area. Secondly, the ADMS working principle is described. In ADMS, the bits are modulated by a high level modulation and coding scheme (MCS), and each DUE can adaptively select one MCS level to demodulate bits according to the channel state information. Finally, the expressions of two performance metrics are derived, namely D2MD system delay and energy consumption, and a distance minimization problem (DMP) with attenuation function to achieve the tradeoff between the two metrics is formulated. From the simulation results, it is confirmed that the proposed schemes improve the performance of the hybrid system compared to the conventional ways. In addition, by controlling the introduced parameter, DMP can balance the tradeoff between the two metrics effectively.

Appendix

Denote the CDF of the received SNR as $F_{\gamma } (\gamma )$. By Eq. 3, the authors get $\left| h \right|^{2} = {{N_{0} \gamma } \mathord{\left/ {\vphantom {{N_{0} \gamma } P}} \right. \kern-0pt} P}$. There must be $F_{{\left| h \right|^{2} }} (x) = P_{r} \{ {{N_{0} \gamma } \mathord{\left/ {\vphantom {{N_{0} \gamma } P}} \right. \kern-0pt} P} \le x\} = P_{r} \{ \gamma \le {{Px} \mathord{\left/ {\vphantom {{Px} {N_{0} }}} \right. \kern-0pt} {N_{0} }}\} = F_{\gamma } ({{Px} \mathord{\left/ {\vphantom {{Px} {N_{0} }}} \right. \kern-0pt} {N_{0} }})$.

Let $t = {{Px} \mathord{\left/ {\vphantom {{Px} {N_{0} }}} \right. \kern-0pt} {N_{0} }}$, and the authors can get $F_{\gamma } (t) = F_{{\left| h \right|^{2} }} ({{N_{0} t} \mathord{\left/ {\vphantom {{N_{0} t} P}} \right. \kern-0pt} P})$. Thus, $p_{{m_{i} }}$ can be given by

$$\begin{aligned} p_{{m_{i} }} = & F_{\gamma } (\gamma_{{m_{i + 1} }} ) - F_{\gamma } (\gamma_{{m_{i} }} ) \\ = & \left\{ {\begin{array}{*{20}l} {F_{{\left| h \right|^{2} }} ({{N_{0} \gamma_{{m_{i + 1} }} } \mathord{\left/ {\vphantom {{N_{0} \gamma_{{m_{i + 1} }} } P}} \right. \kern-0pt} P}) - F_{{\left| h \right|^{2} }} ({{N_{0} \gamma_{{m_{i} }} } \mathord{\left/ {\vphantom {{N_{0} \gamma_{{m_{i} }} } P}} \right. \kern-0pt} P}),} \hfill & {if\,\,{\kern 1pt} m_{i} \ne m_{\left| M \right|} } \hfill \\ {1 - F_{\gamma } (\gamma_{{m_{i} }} ) = 1 - F_{{\left| h \right|^{2} }} ({{N_{0} \gamma_{{m_{i} }} } \mathord{\left/ {\vphantom {{N_{0} \gamma_{{m_{i} }} } P}} \right. \kern-0pt} P}),} \hfill & {if\,\,{\kern 1pt} {\kern 1pt} {\kern 1pt} {\kern 1pt} m_{i} = m_{\left| M \right|} } \hfill \\ \end{array} } \right. \\ \end{aligned}$$