In this paper, we present a novel algorithm for multipath channel estimation for a known preamble sequence in the presence of significant phase noise (PHN) that affects both transmitter and receiver oscillators. The PHN exists in both the receiver and the transmitter and is assumed to vary significantly over the channel length, so it is not possible to approximate the PHN to appear only in one of them. The nonlinearity induced by PHN presents obstacles for conventional linear channel estimation algorithms such as least squares, linear MMSE, or Kalman filters. To address this challenge, we introduce an algorithm based on sequential Monte Carlo methods, also known as Particle Filter. The algorithm’s initial condition is derived using a novel low-complexity windowing method. In addition, we suggest a specially designed preamble sequence that improve further the performance.