Cycle Slip Detection to Provide Stable Precision Positioning under Severe Ionospheric Conditions

Abstract

Carrier phase observations generally have a smaller noise level than code measurements, so they have been used for high precision GNSS positioning techniques such as RTK (Real-Time Kinematic) and PPP (Precise Point Positioning). Recently, the study of autonomous mobility is actively conducted, and high precision GNSS positioning techniques are considered to be a key technology for that. Autonomous mobility requires very high reliability for the safety of life as well as precision. However, the carrier-phase-based positioning system provides a precise navigation solution in normal conditions but failed to detect errors depending on the environment through which the signal passes cause severe error because the carrier phase measurements include the integer ambiguity. Especially, ionospheric disturbance like ionospheric scintillation causes amplitude and phase variation on signals from GNSS satellites and cycle slips caused by loss of lock observed frequently. If the cycle slip cannot be detected properly, positioning is performed using an incorrect integer ambiguity, and which might seriously degrade its performance. Therefore, cycle slip detection techniques are essential for high precision GNSS positioning techniques, and a lot of cycle slip detection methods have been suggested. Many of them are based on assuming an ionospheric delay varies slowly over time, less than 2cm/s. (Park, 2008) However, it is valid only in the mid-latitude regions. During highly active ionosphere periods or in high/low latitude regions, these methods cannot be used because ionospheric variation is far larger than its noise level, they can frequently trigger false alarms and interfere with continuous GNSS operation. Therefore, in this study, we propose RTK-based positioning techniques and cycle slip detection algorithms that can mitigate the effects of cycle slips and false alarms to provide stable precision positioning in high ionospheric activities. First, we adapted the ionosphere-free combination of triple frequency measurements that can eliminate ionospheric error which depends on the inverse square of the frequency in RTK-based positioning. The basic concept of RTK is to reduce and remove errors common to a base station and rover pair. However, if the ionosphere fluctuates rapidly, the ionosphere error is not removed due to local disturbances. Therefore, we use double-differenced ionosphere-free measurements to remove the ionospheric refraction and their theoretical standard deviation is about 0.009m(L1/L2 combination), 0.008m(L1/L5 combination) is suitable for precise positioning. Using this combination makes it impossible to determine the ambiguity because their ambiguities are not integers. Therefore, the performance of this algorithm depends on whether cycle slip is detected or not and the false alarm occurs. Second, we studied to detect cycle slip and to mitigate the effect of false alarm during severe ionospheric conditions. The legacy method to detect cycle slips is time difference of the Geometry-Free (GF) combination. This method is very efficient because it removes geometry term which must be eliminated for use in dynamic systems. However, this method has the risk of generating frequent false alarms due to the rate of change of ionosphere error term during ionospheric scintillation. Therefore, we use the Geometry-Ionosphere-Free (GIF) combination to detect cycle slip by comparing the rate of change of ionosphere error of L1/L2 and L1/L5 geometry free combination using triple frequency measurements. This method is not only applicable regardless of the motion of the GNSS receiver, but also is less affected by ionosphere disturbance. However, the ratio between the integer ambiguity coefficients that constitute the GIF combination, for certain cycle slip pairs, the value is close to zero within the GIF combination, can cause undetectable cycle slip pairs. To mitigate this issue, additional algorithms are introduced. The comparison of doppler and carrier phase detect big jumps that cannot detect by GIF. And the difference of L2 and L5 carrier phase measurements allows the detection cycle slip pairs of L2, L5 signals that GIF cannot detect. The threshold for each linear combination was selected by considering the standard deviation and the required false alarm and missed detection probabilities, and GIF with supplements algorithms can detect all cycle slip pairs except for the identical cycle slip pairs. Identical cycle slip pairs are considered challenging in many cycle slip detection studies. The probability of occurrence of the identical cycle slip on each signal is very slim, but if it is not detected, it can cause errors up to hundreds of meters, which is a very important problem. Therefore, we use the GF cycle slip detection method with GIF methods. As mentioned previously, geometry-free cycle slip detection algorithms cause frequently false alarms during severe ionosphere disturbance. We tried to solve this problem in the positioning process. All measurements passed through this algorithm can be classified into 4 cases are normal condition, cycle slip detection, missed detection, false alarm. Among them, ‘missed detection’ is the case where a pair of cycle slips proportionally to the frequency ratio is detected only by GIF, and ‘false alarm’ is the case where false alarm due to ionospheric error or identical cycle slip pair is detected only by GF. When the measurement is determined to be ‘false alarm’, we cannot distinguish whether this is false alarm or identical cycle slip pairs, but we know cycle slip pairs are 0 or identical. Therefore, to mitigate this issue, we use double-differenced ionosphere-free wide-lane (IFWL) measurements in positioning. The integer ambiguity of the wide-lane combination is the difference between the integer ambiguity of different frequency, the effect of identical cycle slip in positioning is canceled, so the measurements can be used for positioning when ‘false alarm’ occur. The standard deviation of double-differenced IFWL measurements is about 0.644m, and estimating position using only double-differenced IFWL measurements is not suitable for precise positioning. Therefore, we use ‘Ionosphere-Free Wide-lane Switching Mode’ which combines only measurements classified as ‘false alarm’ into double-differenced IFWL combinations to utilize the advantage of the double-differenced IFWL combination that mitigate the effect of identical cycle slip and false alarm. If the user position is estimated well using these algorithms in the epoch, identical cycle slips can estimate accurately because the unknown variable is only one. Therefore, if ‘false alarm’ occurs, cycle slip can estimate accurately, and the position of the following epochs can estimate without performance degradation. Finally, we applied these techniques to the post-processing program to verify our algorithm. First, we implemented using simulated data to artificially insert cycle slip to verify the performance of cycle slip detection and repair in the algorithms. Simulation has shown that these techniques can properly detect and repair inserted cycle slip and provide a stable precision position even after cycle slip occurs. Second, we implemented using GNSS data logged from Alaska Fairbanks Poker Flat Research Range for the 2015/12/20 storm event. As a result, the proposed algorithms perform more stable than using only conventional RTK techniques and Geometry-Free cycle slip detection method and provide cm-level position performance. The proposed algorithms are a robust algorithm that is not affected by severe ionospheric fluctuations, and we confirm that the algorithms estimate position well even when applied to real data. We expect to reliably perform precise positioning anytime, anywhere if we apply more data that collected high/low latitude and advance these algorithms. REFERENCE [1] Park, Byungwoon. Study on reducing temporal and spatial decorrelation effect in GNSS augmentation system : consideration of the correction message standardization. 2008. Seoul University, Ph.D. dissertation