Nowadays, GPS and also GLONASS technologies are already integrated in all relatively advanced mobile phone terminals as a complement to the standard positioning technique based on the multilateration of radio signals between the radio towers of the cellular telephony network and the phone terminal. These technologies, now integrated in the smartphone mass market, enable the existence of a large and still increasing series of positioning based applications and the provision of the so called Location Based Services (LBS) which have been incorporated in the mobile telephony operators' businesses models. The advantage of the GNSS based technology is the higher accuracy, whereas it has the drawback of needing a longer TTFF (Time To First Fix), especially if used in the stand-alone mode. The alternative to this is using assisted GNSS (A-GNSS) positioning systems, which can use information from a network in order to get the TTFF in a quicker way. The development of the A-GNSS technologies was accelerated by the US FCC's 911 mandate, which requires the position of a cell phone to be available to emergency call dispatchers. GNSS assistance techniques are either based on using information to accelerate the satellites acquisition or on having a support server able to more accurately calculate the device position using the GPS receiver provided information. In any case the use of a data connection is needed to contact the assisting GNSS information, and the network provider can charge for it. Out of the two mentioned GNSS assistance schemes, the one we are going to address in this paper is the one based on the improvement of the satellites acquisition strategy. Satellite acquisition can be accelerated by supplying the mobile device with positioning algorithms with precise orbital and time real time or predicted data. The use of real time data entails a heavy load for the mobile device data connection, and thus, for trying to minimize the dependency from the assisted information source and the amount of data to be transmitted/received, we are going to on focus the generation of long-term orbit and time predictions for feeding A-GNSS. A deep knowledge of the OD & TS (Orbit Determination and Time Synchronization) processes is required for successfully carrying out the target task: generating accurate orbit and time predictions for assisting GNSS positioning. Note that standard OD & TS processes have been optimized for generating just a few hours long clock and orbit predictions. These are the kind of predictions required to be fitted into GPS or GLONASS navigation messages, which are quite frequently updated. The authors of this paper accumulate more than 25 years' experience in precise orbit determination and clock synchronization, and they have contributed to the achievement of the amazing accuracy level that can be reached today in the computation of GPS, GLONASS and Galileo orbits, They have been studying in detail, over the last years, the major error sources affecting the clock and orbits computation. Out of these analyses, they have reached a good understanding about how to generate accurate long term orbit and clock predictions, to be used in A-GNSS applications, or to be used for enhancing the GNSS systems robustness and autonomy. The goal of the present work is to show that computing accurate orbits and clock prediction over relatively long time periods is currently feasible. Our target is predicting orbits with accuracies better than 10 meters (RMS) in 20 days and predicting clocks with better than 35 ns (similar to 10.5 m, RMS) accuracy (depending on the type of clock on board the satellite), i.e. about 14.5 m for the combined contribution (RMS). In this paper we are going to analyse the limitations of the usual OD & TS processes when being used for generating long-term orbit and time predictions. By long-term we understand 15 to 20-day-long periods. We will make an assessment on whether it is the orbit or the clock prediction error the one which is limiting the long-term predictions accuracy, nowadays and considering the evolution of the GNSS systems, in particular the improvement of the on board frequency standards. We will analyse the influence of the estimation arc length and will assess the impact of the solar radiation pressure modelling errors in the quality and accuracy of the prediction products. We will also investigate different strategies for the dissemination of the generated predictions and we will measure how the ERPs (Earth Rotation Parameters) accuracy can affect the quality of the predicted orbit and clock products. Ordinary comparison operations will be carried out for assessing the accuracy of the orbit and clock long-term predictions. With all these analyses, we intend to pave the way towards a solid proposal for an OD & TS long-term prediction products generation strategy. A significant part of the derived strategy will be able to be implemented in a short time period, and interesting ideas for the future improvement of the A-GNSS techniques will also be addressed for. A-GNSS based applications and services demand exists and will keep on increasing, and our objective is to take on the challenge of applying our expertise in the OD & TS field with the aim of accomplishing the ambitious task of efficiently providing precise orbit and clock long-term prediction products. Other uses, such as the enhancement of the GNSS systems robustness and autonomy, can also be considered.
