The purpose was to investigate and assess, from a signal processing point-of-view, the impact of the future L5 signal structure on GPS receiver operations from acquisition and tracking to measurements formation and navigation solution derivation. The major conclusions of this work are as follows:
Acquisition Performance
In order to alleviate the high computational load associated with combined algorithms, the acquisition of the PRN and NH codes was performed sequentially.
1. The implementation of the L5 coarse acquisition step is affected by the potential occurrence of unknown NH bit sign transitions. This requires the implementation of zero-padding strategies and constrains the coherent integration time to exactly 1 ms
which, in turns, limits the achievable correlation gain and affects the L5 coarse acquisition sensitivity. To increase the equivalent C/N0 (and, therefore, enhance
detection performance) several data/pilot combining algorithms were proposed and compared. The coherent combining method that makes use of the synchronicity and orthogonality of the data and pilot channels was shown to provide the best theoretical detection performance, followed by the differential and non-coherent combining strategies.
2. The L5 fine acquisition involves the introduction of an intermediate tracking step. This 1-ms FLL-based tracking strategy reduces the frequency uncertainty after coarse acquisition and enables reliable acquisition of the received NH code delay. The NH code delay acquisition is performed using the pilot channel only; this strategy combines robustness with simplicity, and benefits from the superior NH20 correlation
properties.
Constant Bandwidth Tracking Performance
The improvements brought by the L5 pilot channel were confirmed in terms of carrier and code tracking.
1. The main asset, for a carrier tracking loop running on a pilot channel, resides in the use of more efficient discriminators. Since they do not need to address the unknown data bit transition issue, the pilot discriminators used for phase and frequency tracking can provide significant gain in terms of frequency tracking accuracy and sensitivity. These improvements mostly derive from the fact that the pilot discriminators possess extended stability and linear tracking domains. The sensitivity
gain enabled by the pilot discriminators was shown to be approximately 6 dB for both PLL and FLL. From a phase tracking stand-point, the use of a pure PLL discriminator also ensures better resistance to dynamics and oscillator errors, and therefore limits the occurrence of cycle slips. This is of particular interest for carrier smoothing of the pseudoranges and/or carrier ambiguity resolution for double difference carrier phase positioning.
2. Phase tracking and frequency tracking were compared in terms of accuracy and sensitivity. As expected, phase tracking was shown to be the most accurate. Similarly, phase tracking was found to provide higher tracking sensitivity.
3. The pilot channel also allows the use of long coherent integration times. The use of long coherent integration times on the carrier tracking loop is limited by both the quality of the receiver FTS and the expected receiver dynamics. In contrast, the use of long coherent integration times on the code tracking loop is enabled by the carrier aiding that absorbs the effects of oscillator frequency noise and receiver dynamics. This was shown to provide significant accuracy gain, especially at low C/N0.
4. From the above conclusions it would seem reasonable to perform the GPS L5 tracking on the pilot channel. Combined with a single prompt in-phase correlator on the data channel (to enable subframe synchronization and navigation message decoding), this pilot-only tracking strategy would combine robustness and low computational burden. This approach, however, does not make use of all the available power which, in turns, can reduce measurements accuracy. To circumvent this problem a coherent data/pilot combining at the correlator level was introduced. It was shown to provide significant gains in terms of code and carrier tracking accuracy. Its
performance, in terms of tracking sensitivity, was shown to fall halfway between those of the data and pilot channel implementations.
Kalman Filter Based Tracking Performance
The benefits of Kalman filter-based tracking were demonstrated in terms of code and carrier tracking sensitivity and accuracy.
1. The foremost benefit of the KF tracking implementation resides in its modeling capacities. Specifically, the opportunity to include the effects of frontend filtering, oscillator phase noise or dynamics in the measurement and dynamic models effectively provides adaptive bandwidth filtering and reduces the impact of noise on the code and carrier tracking loops.
2. The accuracy improvements are more significant in terms of carrier tracking since the Doppler estimates derived from the KF tracking loop are about one order of magnitude less noisy than those derived from the CB tracking loop. This is of particular interest since it is the carrier tracking loops that show the lowest performance when CB tracking is used.
3. In terms of sensitivity, KF tracking was shown to outperform its CB counterpart by approximately 3 dB when the receiver operates in stable conditions.
4. The performance of Kalman filter-based tracking was also shown to be less dependent on the incoming signal C/N0 than its constant bandwidth counterpart.
However, some accuracy improvements were still observed when the data and pilot channel were coherently combined at the correlator level.
PVT Solution
Using an epoch-by-epoch LSA, the impact of code and carrier tracking accuracy was illustrated at the position and velocity levels.
1. As expected, the advantages of the Kalman Filter-based tracking were confirmed in the position domain. In particular, the accuracy of the velocity estimates obtained when using KF tracking were shown to be approximately one order of magnitude better than those derived from CB tracking. In terms of position accuracy, the improvement was shown to be at the centimetre level.
2. The impact of satellite geometry on the accuracy of the position and velocity estimates was briefly introduced. Specifically, the position and velocity estimate were shown to be noisier in the up direction. Besides, measurement redundancy was shown to enable some accuracy gain in terms of velocity estimation.