Showing papers by "Rahul C. Shah published in 2005"

When does opportunistic routing make sense

Abstract: Different opportunistic routing protocols have been proposed recently for routing in sensor networks. These protocols exploit the redundancy among nodes by using a node that is available for routing at the time of packet transmission. This mitigates the effect of varying channel conditions and duty cycling of nodes that make static selection of routes not viable. However, there is a downside as each hop may provide extremely small progress towards the destination or the signaling overhead for selecting the forwarding node may be too large. In this paper, we provide a systematic performance evaluation, taking into account different node densities, channel qualities and traffic rates to identify the cases when opportunistic routing makes sense. The metrics we use are power consumption at the nodes, average delay suffered by packets and goodput of the protocol. Our baseline for comparison is geographic routing with nodes being duty cycled to conserve energy. The paper also identifies optimal operation points for opportunistic routing that minimizes the power consumption at nodes.

On the performance of geographical routing in the presence of localization errors [ad hoc network applications]

Abstract: In this paper, a detailed study of the performance of geographic routing protocols in the presence of localization errors is carried out. Both analytical and simulation results illustrate the major impact or localization errors on the protocol goodput and route discovery energy. The performance metrics observed were the packet delivery ratio and the power consumed at a node for routing. It is shown that significant performance deterioration occurs with location errors as low as 20% of a node's radio range with no other obstacles in the network. To counteract this degradation, an enhancement is proposed that increases the error tolerance to about 40% radio range and in addition improves the performance consistently for any location error. Furthermore, the effect of obstacles in conjunction with location errors on the routing performance is also investigated.

Modeling and analysis of opportunistic routing in low traffic scenarios

Abstract: Opportunistic routing protocols have been proposed as efficient methods to exploit the high node densities in sensor networks to mitigate the effect of varying channel conditions and non-availability of nodes that power down periodically. They work by integrating the network and data link layers so that they can take a joint decision as to the next hop forwarding node based on its availability and suitability as a forwarder. This cross-layer integration makes it harder to optimize the protocol due to the dependencies among the different components of the protocol stack. In this paper, we provide a framework to model opportunistic routing that breaks up the functionality into three separate components and simplifies analysis. The framework is used to model two variants of opportunistic routing and is shown to match well with simulation results. In addition, using the model for performance analysis yields important guidelines for the future design of such protocols.

Ultra-Low Power Integrated Wireless Nodes for Sensor and Actuator Networks

Abstract: In earlier chapters, it was described how the capability to build and deploy dense wireless networks of heterogeneous nodes collecting and disseminating wide ranges of environmental data enables a multiplicity of exciting scenarios. To mention just a few: smart homes with optimized environmental control and energy management, coordinated media delivery, integrated security, identification and personalization, robot control and guidance in automatic manufacturing environments, warehouse inventory, integrated patient monitoring, diagnostics and drug administration in hospitals, and interactive toys and museums. The mind-boggling opportunities emerging from these technologies indeed give rise to new definitions to the terms “ubiquitous computing” and “user interface”. Regardless of the specific application, however, they all rely on a network of ubiquitously-distributed sensor, compute and actuation nodes, which are integrated and embedded into the fabrics of our daily living environment. This explains why the name “ambient intelligence” is often attributed to such environments. Widespread deployment of these ubiquitous networks requires that some economic and physical realities are met. More precisely, the physical implementation of an individual network node is constrained by three important metrics: cost, size and power. A true ubiquitous deployment is only economically feasible if the cost of the individual elements is ignorable, or, in other words, the electronics have become disposable. Depending upon the intended market and the number of nodes required for a single deployment, this translates to price points per node ranging from $10 to substantially below $1. Achieving a node cost this low requires a minimal number of components, a high level of integration, simple and cheap packaging and assembly, and avoidance of any expensive components and/or technologies. In addition, the cost for the deployment and the maintenance of the network should be ignorable.

Distributed algorithms to maximize the lifetime of wireless sensor networks

Abstract: Maximizing the network lifetime while maintaining application constraints of delay and reliability is the most important goal while designing protocols for wireless sensor networks. This dissertation focuses on a complete protocol stack solution that deals with the problem from two angles. The first part is a routing and MAC layer protocol that minimizes the power consumption required to transmit packets across the network. The cross-layer solution, called region-based opportunistic routing, utilizes the spatial diversity due to high node density to lower the average power consumption, reduce latency and be more robust to bad channels and node failures. Opportunistic routing is a new routing paradigm where the network layer only selects a set of potential forwarding nodes, while the MAC layer does the actual next hop selection based on node availability. This leads to improvements of up to 30% in the average power consumption and 40% in the end to end latency of packets over traditional approaches such as geographic routing. The second part of the solution is a mechanism to shift the forwarding workload among nodes so as to maximize the time till the first node runs out of energy. This is achieved by a distributed duty cycling algorithm that adjusts the duty cycle of each node individually without requiring any communication whatsoever among nodes. The algorithm ensures that the duty cycles of the nodes achieve the optimal duty cycle that minimizes the total power consumption of the network while ensuring fairness in node lifetimes. Moreover, it also ensures that the application latency and reliability constraints are satisfied. The optimal linear control policy is derived, which is a weighted multiplicative increase multiplicative decrease policy. Simulations show that the algorithm achieves a network lifetime fairly close to the optimal lifetime that would have been possible using a centralized approach that had complete knowledge about the network topology, traffic and channel conditions.