Have you considered using a proxy to capture the requests between the phone and the server, assuming for a moment that those requests transit across HTTP?
if it is true that HTTP is in use, setup VUGEN for proxy mode recording. Assuming you can use your phone on the same network as your VUGEN host and their are no intervening security issues, then set your phone HTTP global proxy to the VUGEN proxy IP address and port number.
Start VUGEN. Start recording. Start your application. You should see an HTTP layer conversation being recorded.
Incidentally, as you are trying to compare mobile to desktop, consider some "big data" style log analysis. Every mobile phone includes "mobile" as a key word in the user agent string. Consider finding a request in your production HTTP logs which is common to both mobile and non mobile users. Assuming your a tracking the w3c time-taken field in your logs, then you can compare the statistical information from mobile and non mobile clients to see how long it takes to serve a common component to both client types globally, as well as any given point in time (such as last thursday from noon to 4pm eastern.
When you test for mobile it becomes critical that you model network impairment as the network has at least as great an impact on the client response time as does the application itself in congested 2/3/4g environments.