Comparison and combination of TAI time links with continuous GPS carrier phase results

The time transfer techniques used to generate TAI are currently the TWSTFT (TW), GPS C/A, and GPS P3. About 19% of all TAI laboratories possess P3, 12% possess TW backed up with P3, i.e., in total only 31% of the TAI laboratories transfer 100% of the primary frequency standards (PFS) including all the Cs fountains and 80% of the total weighted clocks for TAI. GPS carrier phase (CP) data are co-products of P3 receivers and they are never used in TAI. An effective strategy to improve TAI is to strengthen these TW+P3 links by using the already available CP information. We first review the classical TAI time transfer techniques and solutions of different analysis strategies of the CP data, namely the IGS [1], the AIUB [2], and the NRCan PPP [3]. To evaluate all the different time transfer techniques, we carried out an extensive comparison between the above techniques, over different periods from hours to 4 months and over different distances on inner- resp. inter-continental baselines between laboratories in Asia, Australia, Europe, and America. We conclude that the CP only solution without using the pseudorange data at all supplies high short term time stability. Comparison over 4 months between all the GPS code supported CP solutions agree perfectly with each other within 0.5 ns and their agreements with TW are of the order of 1 ns or below. However neither TW nor GPS has as a single technique a dominant advantage w.r.t. the others: the TW would be affected by a diurnal disturbance, the CP only by a drift and the P-code solutions by strong noise. The solutions of IGS, AIUB, and PPP are the direct combinations of GPS only Code+CP observations. In fact, the GPS CP-only information can be combined with any absolute time scale [4]. We briefly discussed the concept of the post combination of CP with TW, GPS P3 and C/A code. We studied the methods to combine different techniques which allow maintaining the advantages of a link while reducing its disadvantages. The Vondrak-Cepek method is applied [5]. We prove that the combination of TW and CP greatly reduces the diurnal disturbance in TW Ku band as well as the drift effect in CP only solutions such that the short and long stabilities are both improved. With a careful configuration of the weights and filters, it is possible to combine more than three types of links, like TW, P3, and CP. This allows us to improve both the uB and uA uncertainties in the TAI time transfer. With the help of the high precision TW and CP solutions, we evaluated also the GPS common view (CV) and all in view (AV) time transfer techniques [6]. The latter is considered to replace the common view in the TAI computation.