Ionosphere represents one of the most important challenges and uncertainties in GNSS signal reception. In particular, the scintillation effect, which is a kind of multipath in the ionosphere, produces rapid phase shifts and amplitude perturbations to the GNSS signal. As the next solar maximum is approaching, GNSS receivers must be designed to withstand the unexpected ionospheric disturbances. The paper proposes an approach for the design and implementation of a robust GNSS signal tracking against scintillation. In the literature, for accounting of scintillation, a third-order wide-bandwidth phase-locked loop (PLL) accompanied by cycle-slip detector has been proposed for the carrier phase tracking. However, severe scintillation, which cause deep amplitude fade coincident with fast phase shift, may lead PLL to lose lock. Hence, an approach is needed to keep tracking even in such rigorous environment. The paper adopts the particle filter concepts to realize the carrier signal tracking loop and employs a software receiver to implement the tracking scheme. The particle filter is a sequential Monte Carlo method in which probability is represented in terms of particles. The method allows for a complete characterization of posterior distribution of the estimated state. It is also capable of accounting for nonlinear dynamics and non-Gaussian noise. It is thus expected that the particle filter can potentially overcome the nonlinear amplitude and phase variations due to scintillation, yielding a robust signal tracking. In the GNSS signal tracking particle filter, a dynamic model is used to characterize the amplitude, frequency and phase variations. The I and Q correlator outputs instead of phase and frequency error serve as the measurements. Both process and measurement equations are further subject to noise. The noise covariance is adjusted in accordance with the receiver operating state and noise statistics. To wipe off navigation data bits, a navigation data detector is merged into weight updating of particles. It is noted that although particle filter can be used to address nonlinear dynamics and non-Gaussian noise, the computational complexity may be very high. To lessen the computational load, the software GNSS receiver is employed in the estimation of phase and frequency. By adopting an existing scintillation model, a simulation is conducted to assess the particle filter in the software receiver for GNSS signal tracking in the presence of strong scintillation. In comparison with PLL, the results reveal that the proposed particle filter has better tracking performance under deep amplitude fading and fast phase shifts.