Abstract: Objectives： Global Navigation Satellite Systems (GNSS) are widely used in everyday positioning applications. However, in dense urban canyon environments, severe signal blockage and multipath propagation significantly degrade satellite visibility and observation geometry, leading to large positioning errors. Under such conditions, conventional GNSS positioning solutions often fail to meet the accuracy requirements of demanding applications such as ride-hailing services and pedestrian positioning. To address these challenges, an urban GNSS positioning technique based on receiver clock error modeling and pedestrian spatial constraints is developed to improve pedestrian positioning accuracy and robustness in dense urban environments. Methods： In the temporal domain, a chip-scale atomic clock (CSAC) is employed as a local time-frequency reference. Benefiting from the high stability of CSAC, the receiver clock error is accurately modeled and predicted, and the predicted clock error is incorporated into the GNSS positioning model as a temporal constraint, thereby enhancing positioning stability. In the spatial domain, sidewalk constraints are introduced to control positioning errors in the cross-street direction. By exploiting the asymmetric characteristics of GNSS signal reception on both sides of urban streets and integrating pedestrian sidewalk network data, the sidewalk on which the pedestrian is located is identified. Corresponding sidewalk constraints are then applied to restrict the GNSS positioning solution to feasible pedestrian motion space. The effectiveness of the technique is evaluated using real-world measurement data collected in urban canyon environments. Results： The proposed technique significantly improves positioning accuracy in urban canyon environments. Compared with conventional GNSS positioning solutions, the positioning error is reduced from approximately 20 m to 5-10 m. Compared with the unconstrained solution with an average RMSE of 28.19 m, the CSAC-based clock-plus-height constraint and the sidewalk-plus-height constraint reduced the average RMSE to 5.82 m and 3.69 m, respectively. The incorporation of CSAC-based receiver clock error constraints substantially enhances positioning stability and accuracy. Furthermore, the introduction of sidewalk constraints yields the best positioning performance by effectively suppressing cross-street positioning errors caused by signal blockage and multipath effects. Conclusions： The urban GNSS positioning technique based on receiver clock error modeling and pedestrian spatial constraints effectively improves positioning performance under challenging urban canyon conditions. By exploiting receiver clock stability enabled by chip-scale atomic clock (CSAC) technology and incorporating sidewalk constraints, the proposed approach reduces solution dimensionality, improves robustness, and enhances positioning accuracy without relying on complex environmental models. The results indicate that receiver clock error modeling and pedestrian spatial constraints modeling provides a feasible and effective technical pathway for improving pedestrian GNSS positioning performance in dense urban environments.