In this paper, we consider the problem of power control when nodes are non-homogeneously dispersed in space. In such situations, one seeks to employ per packet power control depending on the source and destination of the packet. This gives rise to a joint problem which involves not only power control but also clustering. We provide three solutions for joint clustering and power control.

Abstract—Transmit power control is a prototypical example of a cross-layer design problem. The transmit power level affects signal quality and, thus, impacts the physical layer, determines the neighboring nodes that can hear the packet and, thus, the network layer affects interference which causes congestion and, thus, affects the transport layer. It is also key to several performance measures such as throughput, delay, and energy consumption. The challenge is to determine where in the architecture the power control problem is to be situated, to determine the appropriate power level by studying its impact on several performance issues, to provide a solution which deals properly with the multiple effects of transmit power control, and finally, to provide a software architecture for realizing the solution. We distill some basic principles on power control, which inform the subsequent design process. We then detail the design of a sequence of increasingly complex protocols, which address the multidimensional ramifications of the power control problem. Many of these protocols have been implemented, and may be the only implementations for power control in a real system. It is hoped that the approach in this paper may also be of use in other topical problems in cross-layer design. Index Terms—Design principles, Linux implementation, power control.

This work explores several system issues regarding the design and implementation of routing protocols for ad-hoc wireless networks. We examine the routing architecture in current operating systems and find it insufficient on several counts, especially for supporting on-demand or reactive routing protocols. Examples include lack of mechanisms for queuing outstanding packets awaiting route discovery and mechanisms for communicating route usage information from kernel to userspace. We propose an architecture and a generic API for any operating system to augment the current routing architecture. Implementing the API may normally require kernel modifications, but we provide an implementation for Linux using only the standard Linux 2.4 kernel facilities. The API is provided as a shared user-space library called the Ad-hoc Support Library (ASL), which uses a small loadable kernel module. To prove the viability of our framework, we provide a full-fledged implementation of the AODV protocol using ASL, and a design for the DSR protocol. Through this study, we also reinforce our belief that it is profoundly important to consider system issues in ad-hoc routing protocol design. 1

In earlier work [1], it was shown that when nodes are uniformly distributed, then, asymptotically as the number of nodes is increased, a common transmit power level is almost optimal with respect to the traffic carrying capacity of the network. In [2] this observation was exploited, and a network layer protocol for power control was developed which ensured convergence to the lowest common power level which ensured connectivity. In this paper, we consider the problem of power control for situations which fall short of the asymptotic regime where nodes may be nonhomogeneously dispersed in space. In such situations, one seeks to employ per packet power control depending on the source and destination of the packet. We address the problem of clustering that results and provide three solutions for joint routing and clustering by power control for ad hoc networks. The first protocol, Clusterpow provides a mechanism for implementing QoS a la Diffserv where power is traded for latency. The second, Tunnelled Clusterpow, allows a finer optimization by using encapsulation. The last, MINPOW, whose basic idea is not new, provides an optimal routing solution with respect to transmit power, but does not readily allow tuning of packet latencies. Our contribution includes a clean implementation of MINPOW at the network layer without any physical layer support. We establish that all three protocols are loop free, while also illustrating how a slightly different approach could lead to packets getting into infinite loops. We provide the software architectural framework of our implementation as a network layer protocol. The architecture works with any routing protocol. Details of the implementation in Linux are also provided.

Traditionally, traffic in wireless networks is routed along minimum hop paths from sources to destinations. This can result in hot spot formation and loss of performance. It can be theoretically shown that flow avoiding routing provides throughput gains over shortest path routing in random wireless networks by a factor of four when the sources are few enough. Motivated by the goal of traffic adaptive routing, we present an alternative approach to wireless network routing using delay measurements to adaptively route packets along multiple paths. The proposed protocol is completely distributed and distributes load along loop-free paths. When the protocol has equilibrated, the network achieves a Wardrop equilibrium, where all utilized paths from a source to destination have the same delay, which is less than that over any unutilized path. We further discuss results from a ns-2 simulation study of the protocol. Our study indicates that the protocol is able to automatically route flows to “avoid” each other, consistently outperforming shortest-path protocols when the number of sources in the network is low. As the number of sources increases, the protocol’s throughput-delay performance is still better than that of shortest-path routing. We also address the architectural challenges confronted in the software implementation of such a multi-path, delay feedback based, probabilistic routing algorithm. A working implementation of the protocol has been built in userspace on a modified Linux 2.4.20 kernel. Finally, we discuss a measurement study of the implementation on a six node testbed.

Routing protocols for multi hop wireless networks have used shortest path routing to obtain best paths to destinations and do not consider traffic load or delay as an factor in the selection of routes. So motivated, we address the issue of designing a multipath routing protocol that is generally applicable even when there are more sources and also we can use link state routing protocol to reduce the impedence. We develop a protocol that adaptively equalizes the mean delay along all utilized routes from a source to destination and does not utilize any routes that have greater mean delay. This is the property satisfied by a system in War drop equilibrium. We also address the architectural challenges confronted in the software implementation of a multipath, delayfeedback-based, probabilistic routing algorithm. Round trip time is taken into account so that we can avoid the duplicate packets from outgoing lines. Age field is included for the sequence number and reduce once per second there will be no packet loss. Heirarchial routing is applied so that it reduces the memory required at each node.