Modeling and forecasting of network traffic data presents a number of challenges in recent paradigm due to the volatility of data. There are various methods used for forecasting time series including AR, MA, ARMA, ARIMA, Fourier transform, ANN, and fuzzy logic. Wavelets technique has attracted the attention of researchers and is a rapidly growing area of research. Using the wavelet transformation, a multiresolution representation of a traffic signal is possible which breaks the signal into its shifted and scaled versions. This breaking up of signal is used for smoothing of time series to differentiate between signal and noise. The filtered and de-noised data is then further used to search time series models as possible candidates for forecasting. These models may be standard AR, ANN or fuzzy, etc to produce forecast that best estimate the mean and variance of actual traffic. Wavelets based Seasonal autoregressive moving average model is used to forecasting network traffic load at University of Karachi’s High Speed Fibre Optics LAN which supports Wireless Computing. The new forecasting strategy incredibly improves the performance.