The use of a network of reference stations instead of a single reference station allows to model some systematic errors in a region, and to increase the operational distance between the rover and reference stations. Permanent GPS reference stations exist in many countries, and GPS observations are available for the users in real-time mode and in post-processing. The paper presents DGPS post-processing positioning with the use of three reference stations. The traditional DGPS technique is based on one reference station. It has been shown that the accuracy of such positioning is about 1-2 meters, depending on the number of satellites being tracked and the resulting value of POOP (Position Dilution of Precision). The accuracy of DGPS positioning degrades when the distance between the rover and the base station increases. The paper shows that when three reference stations are used simultaneously, pseudorange corrections for a virtual reference station, located in the vicinity of an unknown station can be created, and distribution of pseudorange corrections over time can be analysed and modelled. Three reference stations give redundant observations and enable to reduce some measurement errors and biases. Practical calculations and analysis of accuracy have been presented for medium-long and long distances between the rover and reference stations.

This paper describes the prototype version of a mobile application supporting independent movement of the blind. Its objective is to improve the quality of life of visually impaired people, providing them with navigational assistance in urban areas. The authors present the most important modules of the application. The module for precise positioning using DGPS data from the ASG-EUPOS network as well as enhancements of positioning in urban areas, based on the fusion with other types of data sources, are presented. The devices, tools and software for the acquisition and maintenance of dedicated spatial data are also described. The module responsible for navigation with a focus on an algorithms' quality and complexity, as well as the user interface dedicated for the blind are discussed. The system's main advantages over existing solutions are emphasized, current results are described, and plans for future work briefly outlined.