This paper proposes a belief propagation (BP)-based algorithm for sequential detection and estimation of multipath components (MPCs) parameters based on radio signals. Under dynamic channel conditions with moving transmitter and/or receiver, the number of MPCs reflected from visible geometric features, the MPC dispersion parameters (delay, angle, Doppler frequency, etc), and the number of false alarm contributions are unknown and time-varying. We develop a Bayesian model for sequential detection and estimation of MPC dispersion parameters, and represent it by a factor graph enabling the use of BP for efficient computation of the marginal posterior distributions. At each time instance, a snapshot-based channel estimator provides parameter estimates of a set of MPCs which are used as noisy measurements by the proposed BP-based algorithm. It performs joint probabilistic data association, estimation of the time-varying MPC parameters, and the mean number of false alarm measurements by means of the sum-product algorithm rules. The results using synthetic measurements show that the proposed algorithm is able to cope with a high number of false alarm measurements originating from the snapshot-based channel estimator and to sequentially detect and estimate MPCs parameters with very low signal-to-noise ratio (SNR). The performance of the proposed algorithm compares well to existing algorithms for high SNR MPCs, but significantly it outperforms them for medium or low SNR MPCs. In particular, we show that our algorithm outperforms the Kalman enhanced super resolution tracking (KEST) algorithm, a state-of-the-art sequential channel parameters estimation method. Furthermore, results with real radio measurements demonstrate the excellent performance of the algorithm in realistic and challenging scenarios.