No - tracking a proportion of users is perfectly reasonable statistics - done every day in just about every statistical scenario, and with well-established means to estimate the error, based on the size of the sample.
Of course, if the sampling is "biased" then the biases need special consideration, but even that is not impossible or unreasonable. e.g. if 90% of passengers from one area have a smartphone with WiFi switched on, and only 10% from another area of London, then this would need correction, but that's still simpler than manual counting of individual journeys, so by-and-large this seems sensible to me.
Of course there is the "tracking/privacy" issue, but IF done properly, then the WiFi MAC address is a reasonably good, pre-anonymised means to use - certainly less traceable than our faces on CCTV!