In this paper, we propose a new approach in computing interference distribution and outage probability for uplink CDMA cellular networks. In this approach, we model time-varying interference as a discrete-time Markov chain with flexible approximation and find its steady state distribution. Our approach is based on tracing the time variation of the effective factors in interference such as spatial locations, activity status, and shadowing status for the connected users. To this end, we approximate shadowing with several discrete levels by Lloyd algorithm. Moreover, we incorporate correlation properties of the shadowing status in our model. Then, we map the effective factors onto the parameters of a closed Jackson queueing network. With respect to the product-form solution property of Jackson queueing network, we compute the probability distribution of the traffic states corresponding to the various interference statuses. Hence, this approach enables us to find outage probability and following that, static capacity in heterogeneous conditions. At last, we validate the capability of our model for some situations by simulation.