Performance Benefit from Dual-Frequency GNSS-based Aviation Applications under Ionospheric Scintillation: A New Approach to Fading Process Modeling

Markov Chain‐Based Stochastic Modeling of Deep Signal Fading: Availability Assessment of Dual‐Frequency GNSS‐Based Aviation Under Ionospheric Scintillation

“…The normalized histograms of both characteristics were fitted into exponential distributions with their corresponding rates, which were estimated using the SUN ET AL. reciprocals of mean time between fading onsets and fading duration, respectively (Sun et al, 2020). It can be seen that more frequent deep fades (with shorter time between fading onsets) occur in the L5 signal (Figure 4c) than in the L1 signal (Figure 4a).…”

“…The model in Sun et al. (2020) improves on this by additionally modeling fading duration (which also follows an exponential distribution) based on the March 2, 2014 Hong Kong data set. In this method, each fading process is represented by generating starting times of deep fades with the corresponding exponentially distributed random variable of time between fading onsets and independently generating their durations following the fading onsets with the corresponding exponentially distributed random variable of fading duration.…”

“…Time between fading onsets and fading duration are investigated because they are two important characteristics which directly affect navigation availability. Sections 4.1 and 4.2 include a brief review of the Poisson process-based model in Sun et al (2020) to generate fading processes in dual-frequency channels along with a full description of our newly proposed Markov chain-based model. In Section 4.3, the characteristics of deep fades generated from the Poisson process-based model and the Markov chain-based model are compared with those from real scintillation data to evaluate the statistical accuracy of the fading process models.…”

“…Such sequential behaviors of deep fades cannot be described by the superposition of their Poisson processes because each process is independently generated without modeling the correlated temporal structure between them. Due to this restriction, the model in Sun et al (2020) defined new "concurrent" fading events by combining overlapping L1 and L5 deep fades into a single event which spans all of the combined fades, even if only part of the overlapping period includes concurrent deep fading (this is illustrated in Figure 5d). As a result, concurrent fading is common on both L1 and L5 channels.…”

“…However, since scintillation data for GPS L5 was not available at the time, the fading rate of the L5 frequency and the correlation between L1 and L5 frequencies were arbitrarily assumed instead of being based on actual scintillation data. Sun et al (2020) recently proposed an improved model based on statistics of observed GPS L1/L5 scintillation data. However, this model has a significant limitation due to the independent generation of each single-frequency-only and dual-frequency-concurrent fading process.…”