ION GNSS 2009
Session D5: Urban & Indoor Navigation Technology 1

Title: Origin and Compensation of GLONASS Inter-frequency Carrier Phase Biases in GNSS Receivers
Author(s): J-M. Sleewaegen, A. Simsky, W. De Wilde, F. Boon, T. Willems, Septentrio, Belgium
Date/Time: Friday, September 21, 2012, 11:03 a.m.
Room: Chatham Ballroom C

It is well known that centimeter-level inter-frequency biases are characteristic of GLONASS carrier phase measurements. These biases are caused by the use of the FDMA technique (satellites transmit on different center frequencies) and have attracted significant attention in the literature due to their adverse impact on the resolution of integer RTK ambiguities in heterogeneous base-rover combinations. As already postulated by much research, they depend linearly upon the frequency number, change very little with time and temperature, and are almost equal for L1 and L2 bands, when they are expressed in units of length. It has also been shown that the biases tend to be equal for all receivers of a given brand, but significantly differ between manufacturers. Although the general properties of the GLONASS inter-frequency carrier phase biases are well studied, the origin of these biases in the receiver processing chain has remained largely unexplained. It is typically assumed that the biases are generated in the analog hardware of receivers, and hence are difficult to tackle without specialized laboratory equipment.
This paper presents a comprehensive analysis of the possible sources of inter-frequency carrier phase biases in GNSS receivers and proposes a theory of their origin for GLONASS, which explains all actually observed characteristics of the biases, in particular their linearity, stability and equal values for L1 and L2. In a nutshell, the main source of the biases is found in the way phase measurements are generated in the receiver´s digital signal processing (DSP) section. The paper begins with the analysis of inter-frequency carrier phase biases generated in the analog hardware. It is shown that these are too small and cannot account for the observed centimeter-level values. Next, we proceed to the DSP section and identify two mechanisms which may cause the generation of inter-frequency biases. First, it is a common practice to adjust code measurements by some constant offset in the receiver firmware. For example, this adjustment could be performed to compensate for group delay effects in the reception chain in order to align the time at which the pulse-per-second strobe (PPS) is generated. It is shown that such an adjustment, if applied only to code measurements, but not to carrier phase, may lead to significant linear inter-frequency biases in GLONASS carrier phase measurements. The second cause of biases is found in the receiver´s digital signal processing chip. Significant inter-frequency carrier phase biases are caused by different delays of code and carrier phase signals generated locally as reference for the correlation process, the main element of GNSS tracking channels.

These DSP-induced effects, which are major contributors to GLONASS inter-frequency carrier phase biases, are linear with respect to the frequency number and identical on L1 and L2, when measured in the units of length or time. Being caused by digital processing and firmware, these biases are perfectly stable in time, don´t change with temperature, and do not vary from unit to unit. On examples of Septentrio receivers it is demonstrated that the values of the biases, confirmed by independent research, can be fully explained based on the known parameters of the receiver´s DSP.

Being dependent in such a great measure only upon receiver firmware and digital chip architecture, the inter-frequency biases can be compensated in an absolute sense. It is shown that after the compensation is implemented in receiver´s firmware, the inter-frequency carrier phase biases are reduced from centimeter-level values to millimeter-level values, which greatly simplifies the fixing of integer GLONASS ambiguities. The adverse influence of GLONASS phase biases on RTK operation has long been discussed and various proposals to work out unified policy on the subject have been circulated, e.g. the concept of relative inter-receiver calibration by an independent body. Based on the conclusions of this research, a simpler approach can be formulated: each manufacturer may independently identify reasons for biases based on internally available knowledge of applicable parameters of the receiver´s DSP and endeavor to reduce biases to zero (in practice to small mm-level) in an absolute sense. Our approach has been supported by independent researchers during the IGS Workshop on GNSS Biases (Bern, January 18-19, 2012) and further discussions with the interested parties are currently underway.

Previous Abstract Return to Session D5 Next Abstract