Generalized Submodular Optimization for Integrated Networked Sensing Systems
Faculty: Yixin Chen, Chenyang Lu
PhD Student: Abusayeed Saifullah, Chengjie Wu, Paras Tiwari
Alumni: Sangeeta Bhattacharya, You Xu
While wireless sensor networks (WSNs) have traditionally been used as a specialized platform for single applications, recent years have witnessed the emergence of integrated wireless sensor networks as shared infrastructure for multiple applications. The evolution of wireless sensor networks from dedicated platforms to shared infrastructure is driven by a wide range of integrated sensing systems such as urban sensing, building automation, and environmental monitoring. Compared to a separate WSN dedicated to each application, a shared WSN offers more flexibility, adaptivity, and cost- effectiveness through dynamic resource and node allocation.
Due to their resource constraints in bandwidth, memory, and energy, shared sensor networks face a critical need for optimizing the Quality of Monitoring (QoM) through dynamic resource allocation to the contending applications. These emerging QoM optimization problems in shared sensor networks are computationally challenging, as they are discrete and nonlinear. Furthermore, the optimization approach must: 1) deal with multiple resource constraints, 2) scale to large networks, and 3) handle network and environmental dynamics.
To address these challenges, it is important to exploit the special structure of these optimiza- tion problems for sensing applications. A key observation is that most of the QoM functions of the physical phenomena exhibit diminishing returns when more nodes are allocated to an application, a property known as submodularity. Submodular properties are ubiquitous in distributed sensing applications and have been addressed in the literature. However, most of the existing works that utilize submodularity rely on unique problem structures and restrictive assumptions only appli- cable to specific applications. Moreover, all existing submodular optimization algorithms assume a centralized control, which limits their scalability. To ensure scalability and deal with network dynamics, it is essential to develop distributed and online algorithms for submodular optimization. In this project, both centralized and distributed algorithms are proposed to solve submodular optimization problems in WSNs. We develop novel distributed approaches that exploit submodularity. By incorporating submodularity into a market-based framework, we provide the first approximation bounds for distributed submodular optimization.
Faculty: Yixin Chen, Chenyang Lu
PhD Students: Abusayeed Saifullah, Chengjie Wu, Paras Tiwari
Alumni: Sangeeta Bhattacharya, You Xu
While wireless sensor networks (WSNs) have traditionally been used as a specialized platform for single applications, recent years have witnessed the emergence of integrated wireless sensor networks as shared infrastructure for multiple applications. The evolution of wireless sensor networks from dedicated platforms to shared infrastructure is driven by a wide range of integrated sensing systems such as urban sensing, building automation, and environmental monitoring. Compared to a separate WSN dedicated to each application, a shared WSN offers more flexibility, adaptivity, and cost- effectiveness through dynamic resource and node allocation.
Due to their resource constraints in bandwidth, memory, and energy, shared sensor networks face a critical need for optimizing the Quality of Monitoring (QoM) through dynamic resource allocation to the contending applications. These emerging QoM optimization problems in shared sensor networks are computationally challenging, as they are discrete and nonlinear. Furthermore, the optimization approach must: 1) deal with multiple resource constraints, 2) scale to large networks, and 3) handle network and environmental dynamics.
To address these challenges, it is important to exploit the special structure of these optimiza- tion problems for sensing applications. A key observation is that most of the QoM functions of the physical phenomena exhibit diminishing returns when more nodes are allocated to an application, a property known as submodularity. Submodular properties are ubiquitous in distributed sensing applications and have been addressed in the literature. However, most of the existing works that utilize submodularity rely on unique problem structures and restrictive assumptions only appli- cable to specific applications. Moreover, all existing submodular optimization algorithms assume a centralized control, which limits their scalability. To ensure scalability and deal with network dynamics, it is essential to develop distributed and online algorithms for submodular optimization. In this project, both centralized and distributed algorithms are proposed to solve submodular optimization problems in WSNs. We develop novel distributed approaches that exploit submodularity. By incorporating submodularity into a market-based framework, we provide the first approximation bounds for distributed submodular optimization.
Publications
- C. Wu, Y. Xu, Y. Chen, C. Lu; Submodular Game for Distributed Application Allocation in Shared Sensor Networks, The 31st IEEE International Conference on Computer Communications (INFOCOM '12), March 2012. PDF
- A. Saifullah, C. Wu, P. Tiwari, Y. Xu, Y. Fu, C. Lu, Y. Chen; Near Optimal Rate Selection for Wireless Control Systems; The 18th IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS '12), April 2012. PDF
- A. Saifullah, Y. Xu, C. Lu, and Y. Chen; Priority Assignment for Real-time Flows in WirelessHART Networks; The 23rd Euromicro Conference on Real-Time Systems (ECRTS '11), July 2011, pp. 35--44. PDF
- A. Saifullah, Y. Xu, C. Lu, and Y. Chen; End-to-End Delay Analysis for Fixed Priority Scheduling in WirelessHART Networks; The 17th IEEE Real-Time and Embedded Technology and Applications Symposium (RTAS '11), April 2011, pp. 13--22. PDF
- A. Saifullah, Y. Xu, C. Lu, and Y. Chen; Real-time Scheduling for WirelessHART Networks; The 31st IEEE Real-Time Systems Symposium (RTSS '10), November 2010, pp. 150--159. PDF
If you have any questions or comments, feel free to email Chengjie Wu.
Acknowledgements
This work is supported by the NSF under NeTS Grant CNS-1017701.
