Indoor Mapping Based on Virtual Anchors with a Single Base Station

Traditional indoor mapping methods typically rely on multiple base stations (BSs) and sensor fusion, resulting in high costs and limited flexibility. Although single-BS mapping in low-bandwidth, single-input single-output (SISO) networks has attracted significant research interest, it remains challenging. This paper proposes a virtual anchor (VA)-based mapping method using only a single BS, enabling accurate map generation without prior knowledge of indoor geometry. First, we propose an iterative maximum likelihood delay estimation (IMLDE) algorithm that accurately estimates the time of flight (ToF) while reducing computational complexity. Then, a deep neural network (DNN) is employed to model the nonlinear relationship between user locations and VA locations. To refine VA estimates, we introduce an iterative ToF-based selection algorithm. In each iteration, theoretical ToFs for VA-reflected paths are derived under geometric constraints, and the most consistent estimated ToFs are selected to retrain the DNN, progressively improving accuracy. Finally, simulations using commercial-grade wireless ray propagation modeling software show that the proposed method achieves high-precision indoor mapping in resource-constrained scenarios.