Duration of the Thesis: 6
months
Completion: May 2008
Tutor: Dipl.-Ing Doris Becker
Examiner: Prof. Dr.-Ing. Alfred Kleusberg
GPS positioning has been revolutionized with high-sensitive GPS receiver (HSGPS). This receiver variant may track the signal until 20-25dB below conventional receivers can do. Obviously one has no problem when using aircraft or ship, but the problem arises when users loose their position because of signal unavailability in a forest area, down town, and between buildings. Therefore, HSGPS application is started widely used in navigation communities.
When people discuss about HSGPS, suddenly appear questions about its availability and accuracy. Actually, they are always as a center of interest in positioning discipline. Moreover, this thesis discusses about the two factors. Availability of HSGPS in compare with non-HSGPS which gives values not less than 75% of available signal at outdoor area (include shaded location which depends on satellite configuration at the time), but this percentage is decreasing when we observe quantity of carrier phase as 98.06% for good hemisphere and 61.73 for shaded area. In other hand, non-HSGPS gives good availability as good as HSGPS when it is mounted at the good hemisphere area, but when its mounted at the shaded area the percentage significantly decaying 53.41% and 7.23% at indoor which is not able to calculate position.
Nevertheless since HSGPS is a receiver which able to track signal below conventional can do this advantage delivers into accuracy perspective. Here we detect the accuracy level by using three parameters; difference of C/A code with carrier phase in order to see how big the multipath and other disturbances effect occurring, the second method is investigation of carrier phase accuracy by using triple-difference, and the last is comparing signal-to-noise ratio toward elevation. The accuracy of HSGPS is somehow not proportional for the shaded and the indoor area antenna. This expression can be understood since the HSGPS receives all signals no matter direct or indirect.
After analyzing the availability and accuracy from both receivers, processing has been done by using two methods: static and kinematic. They both are based on post processing method which calculating baseline from double-difference method. From both processing systems we compare how big baselines from kinematic method are deviated, in average they give result in decimetre and centimetre fractions.
The downtown environments which contain many buildings and skyscrapers are considered as urban canyon [Gao, 2007]. In perspective signal propagation they are characterized by multipath, signal masking, and lost of signal due to the presence of the skyscrapers and height-rise building. Environmental variables such as height of buildings, orientation of the buildings, material of wall and construction of building, other signals and echoes have role in signal attenuation. If one does measurement under these circumstances, then we can say the condition as 'indoor environment'.
The aim of this thesis is to evaluate the quality of the observables (C/A-code and L1 carrier phase) of the GPS satellites in different situations. The achievable accuracy and availability of positioning has to be performed too. In this case we do experiment of GPS observation in static and kinematic baseline mode. In order to know how good the quality of the observations from C/A-code and L1 carrier phase from each instrument is, we designed some scenarios in which we are working. The scenarios are divided into two main parts:
From two main parts above we derive into real technical aspects by combining them and rearranging into particular scenarios, which are:
The instruments consist of one geodetic receiver and its antenna as fixed station located on the roof; one shared navigation antenna contains two receivers (HSGPS and non-HSGPS) located also on the roof; and another one shared navigation antenna with two receivers (HSGPS and non-HSGPS) as rover. More information about GPS receivers configurations can we see from the following diagram.
|
| Diagram 1. Antenna configuration connected with receivers to design scenarios |
The snr levels are fluctuated depend on how big attenuation factors influence signal propagation. Generally, if we have snr below 30dB-Hz then receiver might not receive carrier phase. Here we can see correlation between snr and signals.
About availability we reorganize percentage table of availability from both outdoor and indoor antenna. Here we only compare availability of HSGPS and non-HSGPS for com10 and com18 since com11 and com19 are not discussed for indoor antenna section.
 |
| Table 1. Availability comparison between outdoor antenna and indoor antenna |
Accuracy of data is distinguished into code-minus-carrier which produces some residues. The residues indicate unsolved ambiguity and some errors. This pre-processing step purposes to see behaviour of satellites signal under such circumstances. From the manner which is showed by individual satellite we can pre-assume how many percents the data under its tolerance and to compare multipath level from the same satellites but different receivers.
Moreover we can see the tendency of satellite by looking connection elevation to snr. The HSGPS at the good hemisphere received the highest snr level, but for the HSGPS at the shaded and indoor area they have lower snr level. This is caused by ability of receiver to receive indirect signals with low snr where the conventional receiver cannot do that. As consequence the HSGPS do not select any data from its multipath level.
Investigation of carrier phase accuracy in order to enhance multipath and noise effects should be supported by rate of range different computation. Without the support the model is still potential with range error which governed by range rate per epoch which generate error propagation of epochs accumulation. Nevertheless this problem can be solved by knowing satellite coordinates of every single epoch which is available from ephemerides, or we can take polynomial to approach the biases.
 |
| Figure 1. Third order polynomial approximation values of multipath and noise level of HSGPS baseline (com11&com10) |
After analyze availability and accuracy of observation raw data, now we discuss about result of every scenario which has been done in the lab.
As we know that we have eight scenarios consist three zero-baseline and five baseline which designed from two antennas plus one geodetic antenna, four receiver (two HSGPS and two non-HSGPS), and three conditions (good hemisphere, shaded area, indoor). All scenarios are processed by using both static baseline and Kinematic method in order to compare the processing quality and which one is more suitable for every case.
 |
| Table 2. Geocentric coordinates and their accuracies for zero-baseline scenarios |
 |
| Table 3. Geocentric coordinates and their accuracies for non-zero baseline scenarios |
From the result above we compare with static baseline method by their baselines calculation as showed by the following table.
 |
| Table 4. Level arms/baseline from all scenarios by using kinematic vs. sttic method, and the different of baseline from both methods |
Some important information are gathered in one processing step and combined depend on focused subject. For availability analysis we used data from elevation and snr over time to see availability code pseudorange and carrier phase availability. From there we extract information that snr, code pseudorange, and carrier phase have threshold which strongly depended on the satellite elevation. To see their interdependence let us have a look the following ideal diagram.
 |
| Diagram 2. Interdependence of snr, C/A-code, and carrier phase to elevation |
We classify elevation into two groups:
This experiment is as a first step of the whole experiment packet about availability and accuracy HSGPS compare with non-HSGPS and also to analyze the best way to process the data. Advanced accuracy detection can be done for post-processing one could determine multipath effect and model cycle slip effect.
Many things in this thesis have been not covered yet due to time limit. Discussion about cycle slip, multipath, and ambiguity solution method for kinematic algorithm are some interesting in a big part that should have discussed. However one can continue into deeper and more analytic way with this topic.