Optimal Cross-Layer Design of Wireless Fading Multi-Hop Networks

The last decade has brought a rapid growth in demand for fast and error-resilient telecommunication services. In accordance with this growth, broadband wireless networks have become an integral part of the global communication infrastructure. Provisioning of quality-of-service (QoS) in broadband wireless networks requires coping with the challenges brought by the wireless interface, and allocating resources available at different layers among nodes, links, and end-to-end connections. Nonlinear optimization tools have been successfully adopted to analyze and design algorithms that fulfill such requirements; see e.g., [12, 18, 35, 53] and references therein. Optimal network designs are obtained by formulating a constrained optimization problem involving variables from different layers, and by exploiting information about the wireless channel. Solving such optimization problems dictates how resources are allocated across different layers, while network control protocols follow from the algorithms used for this solution. One of the most challenging issues to cope with in designing optimal cross-layer resource allocation schemes for wireless networks is the presence of fading. Fading renders wireless channels random, degrades the communication performance, and leads to location-dependent and time-varying link capacities. As a result, cross-layer schemes are required to account for the fading nature of the channel, and implement mechanisms to deal with it. Such schemes should be able to effectively exploit the diversity provided by the channel, and adapt the resource allocation to the channel state information (CSI) available. Capitalizing on optimization theory and stochastic approximation tools, this chapter deals with channel-adaptive algorithms that allocate resources at transport, network, link, and physical layers. These algorithms emerge from the solution of constrained optimization problems that take into account the QoS, the interaction among layers, and the CSI available. The model describing the multi-hop wireless network as well as its most relevant operating conditions are as follows. Nodes receive packets from the application layer intended for different destinations. Flow control and routing decisions respond to packet arrivals, and the long-term average end-to-end rates entail different utility levels. At the link layer, two different models are considered, whereby nodes access either orthogonally or non-orthogonally a set of parallel flat fading channels. Orthogonal here means that if a terminal is transmitting, no other link interfering with this transmission can be active;