This project seeks to investigate architectural design approaches to enable pervasive community-shared sensor grids composed of contributed sensor nodes from multiple sensor networks across different administrative domains and ownership. Two fundamental components of this research which allow controlled sharing of sensor nodes? hardware resources and software programmability are: (1) a low-overhead virtualization of the nodes themselves, including layer-2 bridging support and the ability to switch the mote network context among multiple application entities to form transient -underlays- that connect motes across multiple domains, and (2) a P2P overlay network of the sensonet-Internet gateways that is responsible for dealing with authentication, authorization, formation and teardown of these underlays. The expected outcome of the research is set of network applications that will enable Internet users to contribute their sensor hardware and to carve their own sensor network from these shared resources in a GENI-style sharing. This has the potential for promoting participatory sensing among ordinary citizens, for playing a role in urban sensing and emergency response system in the future, and for influencing the development and commercialization of embedded sensing platforms. A more short-term impact during the course of the project is recruitment and retention of students by coupling class projects with experimentation on design prototypes, especially targeted towards engaging students from under-represented groups into research.

Project Report

This project addresses a key facet of the oncoming vision of the Internet of Things (IoT). IoT is the next generation of embedded devices, controllers, sensors, widgets and appliances, all connected via wired and wireless access networks to the global Internet. These devices are to interact amongst themselves and user portals on computing devices to run the applications of the future. One key aspect of IoT is the ability to form communities out of devices coming from diverse ownership, or administrative domains as per the Internet terminology. Sensors deployed by individuals, communities or organizations, are capable of generating data with varying levels of privacy needs, which may or may not be amenable to sharing among these community applications. We enable a platform of community sensing where a virtual network domain is created on demand out of sharable sensor nodes contributed by different owners, for the sole purpose of temporarily hosting an IoT application and matching its sharing needs. Our work provides the programmers with a single virtual network abstraction covering the physical phenomenon under interest rather than exposing them to the intricacies of network connectivity and efficient routing. The realization of the virtual wireless network abstraction is completely software defined, implemented by novel use of both underlay and overlay networking techniques. The underlay network provides efficient wireless mesh connectivity and coverage with sensor nodes that are geographically close, but not topologically close due to ownership and administrative restrictions. The overlay network is the key to the Internet integration and network management, and provides routing paths and tunnels over the Internet, making even physically distant nodes appear to be close. The virtual network management service, which configures the ephemeral network using an overlay of peer to peer control points, can be hosted on a cloud service. The discovery of hosting networks, the selection of control points and the peering management systems integrate well with social networks, so that usage of social networking such as Facebook or Google+ facilitate our publicity and discovery functions. The intellectual merit of the community sensor grid project is in the conceptualization of the virtual sensing platform, the community formation ideas, the virtualization of the hosting network substrates and the innovative methods used in realizing the design without significant increase in power consumption or sensor node deployment. The power saving is achieved by reducing the semantic gap between the virtual network abstraction and the reconfigurable physical networks. The project has significant broader impacts on society, technology and business. Community sensor grids are expected to be major enabler of IoT applications. The underlying growth of IoT itself would benefit society both via technological progress and pervasive sensor coverage to improve quality of life. Integration of social networks makes usage easier and brings technological progress to common participants. Businesses that provide services, applications, advertising and other services will benefit from harnessing such ideas and techniques.

Agency
National Science Foundation (NSF)
Institute
Division of Computer and Network Systems (CNS)
Type
Standard Grant (Standard)
Application #
1011931
Program Officer
Thyagarajan Nandagopal
Project Start
Project End
Budget Start
2010-01-15
Budget End
2013-12-31
Support Year
Fiscal Year
2010
Total Cost
$331,735
Indirect Cost
Name
Arizona State University
Department
Type
DUNS #
City
Tempe
State
AZ
Country
United States
Zip Code
85281