bmwcarit
Using time difference in computation of transmission probabilities
The Newson and Krum paper uses route distance differences to calculate transmission probabilities. Why is time difference being used in this implementation ?
The probability model used here addresses a similar problem as described here: https://github.com/bmwcarit/barefoot/issues/135#issuecomment-471323764 The solution here, however, is different and does not address pitfall B. EDIT: And, of course, the inclusion of the time parameter is used to make it more robust to varying sampling rates: https://github.com/bmwcarit/offline-map-matching/blob/9e03254591ce71c108303788c962203a85210e5c/src/main/java/com/bmw/mapmatchingutils/HmmProbabilities.java#L82-L84 (In Newson and Krumm, different time intervals between subsequent measurements would require different values for parameter beta.) To solve the same problem also other solutions are possible as shown for example in the referenced discussion.
