Joint Channel Estimation and Symbol Detection for Pilot-Free Superposition Transmissions Over Dynamic 3D Trellis

Superposition transmissions can simultaneously accommodate multiple users, while incorporating pilot-free design can further reduce the signaling overhead. Achieving reliable communication for pilot-free superposition transmissions faces previously unaddressed challenges not only due to unignorable inter-user interference (IUI) but also to unknown channel state information (CSI). Considering such difficulties, we first propose a dynamic 3D trellis diagram to effectively characterize the pilot-free superimposed signals, where IUI is introduced as an additional dimension, and a notion of updatable path metric is proposed to adapt to instantaneous CSI estimation. An interference-aware forward-backward Viterbi algorithm is then proposed for pilot-free joint channel estimation and symbol detection, with forward path selection using Gaussian mixture model-based piecewise metric calculation and backward CSI update using survivor path metric maximization. Interference reconstruction and cancellation are further developed for iterative interference suppression among users. The convergence of the proposed algorithm is also theoretically proved. Simulation results indicate that our method achieves an order-of-magnitude frame error rate improvement over state-of-the-art methods.