Adaptive resource management for DS-CDMA networks subject to energy constraints

We consider a time-slotted DS-CDMA network consisting of a single radio access point and a finite number of mobile wireless terminals. The terminals generate non-real-time packet data and transmit, subject to constraints on the average energy consumed per packet transmission, in a random access fashion over a common broadband channel to the access point using different spreading codes. As in narrowband ALOHA systems, the performance of random access spread spectrum networks can be severely hampered due to saturation effects caused by inherent bistable behavior. We study the effect of dynamic spreading gain control on the stability and throughput properties of the network. We first impose no energy constraints and provide an optimal (throughput maximizing) algorithm under which (i) the asymptotically optimal retransmission probability is equal to one, and (ii) the optimal spreading gain increases linearly as the multiaccess interference (MAI) level increases, or equivalently, the transmission rate decreases inverse linearly as the MAI level increases. We also obtain a simple closed-form expression for the asymptotic throughput as the number of backlogged terminals becomes large. We then impose energy constraints and modify the optimal algorithm by controlling both the spreading gain and retransmission probability of the terminals. The resulting algorithm may or may not eliminate bistability; however, in either case, it achieves high throughput.