Downlink resource allocation for the high-speed train and local users in OFDMA systems

We consider providing services for passengers in a high-speed train and local users (quasi-static users) in a single OFDMA system. For the train, we apply a two-hop architecture, under which, passengers communicate with base stations (BSs) via a mobile relay (MR) installed in the train cabin. With this architecture, all passengers in the train can be represented by the MR. Since the channels of the MR and local users vary differently, we consider allocating system resources (power and subcarriers) over two time-scales for them. We formulate the problem as a capacity optimization problem for the MR subject to the sum capacity constraint of local users. We treat the inter-carrier interference (ICI) at the MR as additive Gaussian noise and derive an explicit expression for the ICI using the two-path Doppler spread model. Then we discuss the optimization problem and propose an optimal power and subcarrier allocation (OPSA) policy. The capacity obtained using OPSA is compared with that of constant power and subcarrier allocation (CPSA) policies. Simulation results justify the optimality of the OPSA. Besides, by comparing the capacity bounds achieved by OPSA with and without ICI, we find that only in specific regions, where the gap between the capacity bounds is large, do practical ICI cancellation methods provide meaningful rate gain.