Radio resource management in infrastructure-based and ad hoc UWB networks

Modeling the resource manager of an ultra wide band (UWB) network is the main object of this paper. The model is tested in two different application scenarios: (i) a UWB WLAN access network to a backbone where the resource management module is implemented at the access points (APs); (ii) a UWB ad hoc network for either local communications or data exchange among sensors, with peer-to-peer links with distributed management. The design must include a quality of service (QoS)-aware strategy and must take into account coexistence issues raised by the use of UWB at the physical layer. Link quality is represented by the maximum end-to-end delay and minimum percentage of correct packets. From these parameters, the resource manager jointly selects the values of power and rate which must be adopted at the physical layer. In the model, radiated power by each device is supposed to be limited by an upper bound, reflecting thus the limitation imposed by regulation. QoS awareness and power constraints are satisfied, thanks to the implementation of an admission control function, which is centralized in the AP in the WLAN case and distributed in the ad hoc case. Major innovative aspects include: (i) taking into account UWB specific features; (ii) introducing QoS awareness based on network layer parameters rather than physical layer parameters; (iii) incorporating error protection functionalities for the optimization of transmission efficiency; (iv) considering both centralized and distributed resource management. Performance of the proposed resource manager module is evaluated in the presence of different classes of traffic, that is multimedia, voice, and data traffic. For each class of traffic, performance is expressed in terms of the maximum possible number of simultaneous connections for the WLAN case, and in terms of the effective achievable throughput for the ad hoc case. Results of simulations indicate that the behavior of the proposed scheme is strongly dependent upon the class of traffic in the WLAN scenario, while it is slightly affected by changes in traffic characteristics in the ad hoc case. Copyright (c) 2005 John Wiley & Sons, Ltd.