Research finds within the last decade on network measurement have shown the classical assumption of exponentially decaying service distributions does not apply in a number of sources - Ethernet, ISDN, CCSN and VBR video. Recent research has shown that self-similar nature of traffic still exists in the traffic of wireless networks. Appropriate service distribution of these sources may be heavy-tailed service distributions. A difficult with heavy-tailed service distribution queuing model is that they may not have closed-from, analytic results. This paper proposes a method for analyzing the self-similar CMWAP traffic via HErd (Hyper-Erlang distribution) fitting technique and matrix geometry method. Analytical results and Numerical results of queuing performance effectively describe the self-similar nature of wireless network node against traditional Poisson process.