I'm looking for implementations of privacy-preserving shortest path algorithms that offer real-time or near real-time query performance, where both the source and destination nodes are kept secret. I've explored a few options but haven't found anything that meets the performance requirements:
- The approach described in "Privacy-Preserving Shortest Path Computation" (https://arxiv.org/pdf/1601.02281) seems promising, but I couldn't find any available implementation.
- I've also looked at the Dijkstra implementation in MP-SPDZ (https://github.com/data61/MP-SPDZ/blob/master/Programs/Source/dijkstra_example.mpc), which is based on https://www.iacr.org/archive/asiacrypt2014/88730241/88730241.pdf. However, this implementation is quite slow for real-time applications. I've considered modifying the MP-SPDZ implementation to use A* instead of Dijkstra's algorithm, but this seems challenging due to the need for branching on secret data which is not possible.
Are there any other privacy-preserving shortest path implementations or approaches that could offer real-time or near real-time performance? I'm particularly interested in solutions that have been already implemented. Maybe some homomorphic encryption based implementions?
