Nihar Jindal, Ph.D. - US grants

Title: Collaborative Research: Cognitive Ad Hoc Networks: Capacity Optimization Through Local Adaptation

0635003, Weber
0634979, Andrews
0634763, Jindal

Due to the unpredictability of the environment in which unplanned (ad hoc) wireless networks will operate, an appealing approach is to allow the network to dynamically adapt to the perceived conditions. We define such ad hoc networks as cognitive. A framework is developed for understanding the benefits of local adaptation, by breaking adaptive techniques into the four major degrees of freedom available to the designer: time, frequency, code, and space. The aim is to address the following two questions. First, what are the fundamental limits on information flow through unplanned networks; in particular, how valuable is localized information and coordination in seeking to achieve this limit? Second, what are the relative values of adaptation in time, space, frequency, and code in terms of information flow; in particular, how does the network designer identify which degree of freedom is most valuable in a variety of networking scenarios?

In this research, information theory and stochastic geometry are connected through a novel metric for ad hoc network capacity, termed the transmission capacity. This metric captures the maximum spatial intensity of transmissions subject to a specified outage probability. While related to other popular capacity metrics, notably the transport capacity, the transmission capacity is unique in its allowance of explicit and accurate characterization of capacity for any conceivable communication scheme or transmission environment. The transmission capacity is an indispensable unifying metric for this analysis since i) it allows closed-form results, ii) does not require any global coordination or optimization, iii) accurately models the interference environment of an ad hoc network. These innovations will allow a basis for more efficient wireless network design.

Hondzo
0607138

The water quality of streams draining watersheds has been degraded by increasing urbanization. The general symptoms of this degradation include more frequent large flow events, reduction in channel complexity, reduced retention of natural organic matter, and elevated concentrations of nutrients. Newly emerging urban water quality threats, including insecticides, herbicides, pharmaceuticals, and estrogens, are known or suspected to damage the health of humans and ecosystems. The restoration and management of streams have traditionally attempted to improve the hydrological and water quality conditions in-stream or in riparian zones. Recent studies have indicated the portion of a watershed covered by impervious surfaces and connected to the
stream by stormwater drainage is the primary degrading process of stream ecology and health. These findings suggest that the sustainable restoration and management of stream water quality require quantification of hydrological, chemical, biological, and geomorphological processes, and that these processes must be assessed across a range of scales. Furthermore, interactions among biogeochemical processes across watersheds are either non-linear processes or linear processes dependent on non-linear drivers. The monitoring of such a system inherently requires a change in traditional field sampling strategies. We propose to transform traditional and very limited (in terms of spatial and temporal resolution) field measurements through the integration of multi-scale, spatially-dense, high frequency, real-time, and event-driven observations by a wireless network with embedded networked sensing. Intellectual merit: The objective of our research is to establish a wireless network with embedded sensing capable of monitoring fundamental water quality parameters. Such a network is a key component for watershed observatory networks. The ability of these fundamental water quality parameters to be used for predicting the presence of emerging chemical contaminants in urban streams will also be determined. It is hypothesized that the concentrations of emerging contaminants will correlate with the fundamental parameters measured using the sensor network and that the sensor network will give improved prediction of the loads of these contaminants compared to traditional, discrete grab sampling. Our overall hypothesis is that water quality in streams draining impervious areas of urban land is controlled by the mean and variance of effective
stormwater residence time. The mean and variance of water residence time, the time it takes urban runoff to travel between the impervious urban land and a receiving aquatic body, will be quantified by radio frequency identification technology (RFID), tracer studies, and fluid-flow velocity measurements within the proposed wireless network. A small urban watershed will be equipped with wireless networked sensing to address the following objectives: (1) measurement of fundamental water quality and hydrologic parameters with spatiallydense and high frequency resolution, (2) correlation of general parameters with the presence and/or levels of emerging contaminants, and (3) integration of field measurements to the watershed using primarily the mean
and variance of effective stormwater residence time. Water quality in streams will be observable as a dynamic response to land use gradients and hydrological transients rather than as an equilibrium described byaverage properties. This approach will enable process-based scaling and forecasting of water quality in streams from the in-stream processes to the watershed level.
Broader impact: Wireless networks with embedded networked sensing are designed to quantify spatial and temporal heterogeneities of variables across a variety of scales and are perfectly suited not only for water quality management in streams but also to all environmental processes where interconnectivity among different parts determines the overall state of the system. The network to be developed in this project will focus on an urban stream, but can be expanded to include other watersheds with different land uses in the future. Generated data and scaling relationships will transform urban planning practices and management of water quality in streams draining urban land. Rather than focusing on manipulating in-stream processing only,
a more sustainable approach would be to focus on selection and location of stormwater best management practices (e.g., detention ponds or wetlands). The proposed field measurements are focused on evaluating best management practices for more appropriate and effective hydrologic management. The project will have two educational components. One addresses students and scientists through a revised educational curriculum. The
other is an international collaboration with the Technical University of Denmark through a developed Internet
course on "Integrated Urban Water Quality Management."

Wireless networks now constitute a critical component of the national communications infrastructure, and information theory has served as the guiding light for the growth of wireless. However, the critical role that space, i.e., the physical locations of devices, plays in wireless networks has been largely downplayed in wireless research. Indeed, the wireless revolution would not have been possible without spatial reuse (i.e., reuse of spectrum at physically separated locations). If a link or system is studied while ignoring reuse, design conclusions can be vastly different than when reuse is considered. As wireless usage continues to grow, it is becoming increasingly important that a fundamental understanding of the limits and optimal design of networks be developed.

This research introduces a framework for modeling spatial interactions in wireless networks and develops methodologies for analyzing such networks. Spectrum sharing is perhaps the most basic wireless network:
transmit-receive pairs wish to communicate with each other but not with other pairs, but are forced into a shared environment because they must use common spectrum. A canonical model that explicitly models node locations, dubbed the spatial interference channel, is introduced.
Different communication techniques are analyzed and fundamental bounds on information flow are derived. Ad hoc networks are a more sophisticated type of network in which nodes potentially work together to aid communication. The role of space has been acknowledged in this setting, but many theoretical questions remain unanswered. Generalized scaling laws that capture the interplay between network size, throughput, power, and bandwidth are studied, and spatial models are used to quantify the benefit of techniques such as relaying, network coding, and power control.