Continuous monitoring of performances becomes necessary to detect critical situations in the radio interface of digital wireless telephone networks. This paper presents a measurement tool that analyzes TCP-level packet traffic and detects the effects of variations in the radio channel conditions of a mobile network. The analysis algorithm considers the energies of wavelet coefficients of the aggregated traffic process at a suitably selected number of time scales. Wavelet energy variations are detected by a statistical test from which a score is calculated. This indicator reveals changes in the performances of the monitored network.