The development of a dynamic model for a multichannel optical fiber local area network with finite size source buffers based on the Slotted ALOHNALOHA protocol is presented. The input, measured in terms of the number of user requests in each time slot, is modeled as a Poisson random process with time varying mean intensity thereby allowing time varying data traffic. A cost functional reflecting the measure of performance of the system in terms of expected average throughput channel under utilization, and waiting time is introduced. The model allows estimation of design parameters that cannot be achieved by static analysis.