| We jointly optimize the physical (PHY) and medium access control (MAC) layers for enabling rapid, contention-based wireless medium access. The key new features of our cross-layer design are multipacket reception (MPR) capability and the ability to estimate the number of contending users, regardless of success or failure. Users contend by randomly choosing contention slots for transmitting, randomly choosing the "phase", of a common periodic training sequence known to the access point (AP), and randomly choosing short spreading sequences not known to the AP. Adaptive multiuser detection based on the Differential Minimum Mean Squared Error (DMMSE) criterion is employed, thus permitting reliable demodulation of an unknown number of contending users in a contention slot, despite multiuser interference (and possibly a near-far problem), fading, lack of carrier synchronization, and lack of knowledge of the users' spreading sequences. We show that even without coordination between the mobiles and the AP, multiple simultaneous contentions are successful in a contention frame with high probability, thus drastically reducing delay compared to contention based on classical, narrowband ALOHA. We characterize the throughput of the system and design dynamic stabilization policies based on estimates of the backlog, which are generated by extending Rivest's pseudo-Bayesian technique for classical Aloha to exploit the new information from the PHY on the number of transmissions.; The application focus of our cross-layer design is towards enabling rapid, mobile-centric, handoffs in pseudocellular networks with small AP coverage areas as in Wireless Local Area Network technology, but supporting real-time applications under vehicular mobility as in cellular networks. To develop a design framework, we provide an analytical model of the cross-layer design that includes the new MPR and multiplicity feedback features and apply it to two classes of users: delay-constrained, Hi-priority users; and delay-tolerant, Lo-priority users whose throughput we wish to maximize, while guaranteeing QoS for the Hi-priority users. The channel throughput and the achievable QoS are characterized as functions of the arrival rates for Hi- and Lo-priority users and we obtain contention policies that ensure QoS and stability. Finally, we apply these methods to simulations of the DMMSE-based cross-layer design, and show that the analytical model provides accurate guidelines for design and performance predictions. |
