GPS navigation in urban environments is prone to error sources such as multipath and signal blockage. Shadow matching and likelihood-based 3D-mapping-aided (3DMA) ranging methods have shown high potential in 3DMA GNSS navigation for a single agent in urban environments. However, they suffer from errors such as ambiguity which cannot be reliably mitigated without external sensing and integrated systems. Collaborative localization (CL) is a way to aid navigation in a multi-agent system and is capable of providing the external sensing required to reduce error due to ambiguity in 3DMA approaches. CL algorithms face challenges such as scalability, robustness to noisy sensor data and single point of failure, and operability despite limited inter-agent communication. In this paper, we present a decentralized collaborative localization algorithm which is applicable to sparsely communicating networks, and has limited information exchange. Moreover, the proposed algorithm takes advantage of the variable visibility of the sky and variable building geometry for different agents. We propose a methodology for ambiguity mitigation by constraining an agent's probability distribution using its neighbors. The methodology is based on coupling of agents' GPS measurements with range-only sensors and is applicable to multi-agent systems with these modalities. The proposed method is validated on simulated datasets in an urban area of Champaign, Illinois with multiple agents in a variety of scenarios. We demonstrate the improved performance in terms of positioning accuracy and ambiguity mitigation. We also analyze the impact of network connectivity, and size of network on positioning accuracy.