DISTIN: Distributed Inference and Optimization in WSNs A Message-Passing Perspective SCOM Team

Slides:



Advertisements
Similar presentations
Bayesian Belief Propagation
Advertisements

1 A Real-Time Communication Framework for Wireless Sensor-Actuator Networks Edith C.H. Ngai 1, Michael R. Lyu 1, and Jiangchuan Liu 2 1 Department of Computer.
CLUSTERING IN WIRELESS SENSOR NETWORKS B Y K ALYAN S ASIDHAR.
Infocom'04Ossama Younis, Purdue University1 Distributed Clustering in Ad-hoc Sensor Networks: A Hybrid, Energy-Efficient Approach Ossama Younis and Sonia.
Introduction to Wireless Sensor Networks
Presented By- Sayandeep Mitra TH SEMESTER Sensor Networks(CS 704D) Assignment.
A novel Energy-Efficient and Distance- based Clustering approach for Wireless Sensor Networks M. Mehdi Afsar, Mohammad-H. Tayarani-N.
EE462 MLCV Lecture Introduction of Graphical Models Markov Random Fields Segmentation Tae-Kyun Kim 1.
CISC October Goals for today: Foster’s parallel algorithm design –Partitioning –Task dependency graph Granularity Concurrency Collective communication.
Network Correlated Data Gathering With Explicit Communication: NP- Completeness and Algorithms R˘azvan Cristescu, Member, IEEE, Baltasar Beferull-Lozano,
Practical Belief Propagation in Wireless Sensor Networks Bracha Hod Based on a joint work with: Danny Dolev, Tal Anker and Danny Bickson The Hebrew University.
Distributed Regression: an Efficient Framework for Modeling Sensor Network Data Carlos Guestrin Peter Bodik Romain Thibaux Mark Paskin Samuel Madden.
An Efficient Clustering-based Heuristic for Data Gathering and Aggregation in Sensor Networks Wireless Communications and Networking (WCNC 2003). IEEE,
Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display. Parallel Programming in C with MPI and OpenMP Michael J. Quinn.
1 Distributed Online Simultaneous Fault Detection for Multiple Sensors Ram Rajagopal, Xuanlong Nguyen, Sinem Ergen, Pravin Varaiya EECS, University of.
Probabilistic Data Aggregation Ling Huang, Ben Zhao, Anthony Joseph Sahara Retreat January, 2004.
© 2005, it - instituto de telecomunicações. Todos os direitos reservados. Gerhard Maierbacher Scalable Coding Solutions for Wireless Sensor Networks IT.
LPT for Data Aggregation in Wireless Sensor networks Marc Lee and Vincent W.S Wong Department of Electrical and Computer Engineering, University of British.
Mobile Agents in Wireless Sensor Networks Ivan Vukasinovic Zoran Babovic Goran Rakocevic.
An Energy Efficient WSN for a Video Surveillance System (Timeline for Literature Review) By Pakpoom Hoyingcharoen 3 November 2008 (2/2551)
Decentralised Coordination of Mobile Sensors School of Electronics and Computer Science University of Southampton Ruben Stranders,
Sharanya Eswaran, Penn State University Matthew Johnson, CUNY Archan Misra, Telcordia Technolgoies Thomas La Porta, Penn State University Utility-driven.
1 Chalermek Intanagonwiwat (USC/ISI) Ramesh Govindan (USC/ISI) Deborah Estrin (USC/ISI and UCLA) DARPA Sponsored SCADDS project Directed Diffusion
LPT for Data Aggregation in Wireless Sensor Networks Marc Lee and Vincent W.S. Wong Department of Electrical and Computer Engineering, University of British.
Summary Alan S. Willsky SensorWeb MURI Review Meeting September 22, 2003.
Decentralized Scattering of Wake-up Times in Wireless Sensor Networks Amy L. Murphy ITC-IRST, Trento, Italy joint work with Alessandro Giusti, Politecnico.
Energy Efficient Routing and Self-Configuring Networks Stephen B. Wicker Bart Selman Terrence L. Fine Carla Gomes Bhaskar KrishnamachariDepartment of CS.
The Minimal Communication Cost of Gathering Correlated Data over Sensor Networks EL 736 Final Project Bo Zhang.
IPCCC’111 Assessing the Comparative Effectiveness of Map Construction Protocols in Wireless Sensor Networks Abdelmajid Khelil, Hanbin Chang, Neeraj Suri.
Dynamic Clustering for Acoustic Target Tracking in Wireless Sensor Network Wei-Peng Chen, Jennifer C. Hou, Lui Sha.
NSF Critical Infrastructures Workshop Nov , 2006 Kannan Ramchandran University of California at Berkeley Current research interests related to workshop.
Low-Power Wireless Sensor Networks
Adaptive CSMA under the SINR Model: Fast convergence using the Bethe Approximation Krishna Jagannathan IIT Madras (Joint work with) Peruru Subrahmanya.
Network Kernel Architectures and Implementation ( ) Network Architecture Chaiporn Jaikaeo Department of Computer Engineering.
Department of Computer Science City University of Hong Kong Department of Computer Science City University of Hong Kong 1 Continuous Residual Energy Monitoring.
Chapter 3 Parallel Algorithm Design. Outline Task/channel model Task/channel model Algorithm design methodology Algorithm design methodology Case studies.
Message-Passing for Wireless Scheduling: an Experimental Study Paolo Giaccone (Politecnico di Torino) Devavrat Shah (MIT) ICCCN 2010 – Zurich August 2.
Self‐Organising Sensors for Wide Area Surveillance using the Max‐Sum Algorithm Alex Rogers and Nick Jennings School of Electronics and Computer Science.
COMPUTING AGGREGATES FOR MONITORING WIRELESS SENSOR NETWORKS Jerry Zhao, Ramesh Govindan, Deborah Estrin Presented by Hiren Shah.
Minimum Average Routing Path Clustering Problem in Multi-hop 2-D Underwater Sensor Networks Presented By Donghyun Kim Data Communication and Data Management.
Lan F.Akyildiz,Weilian Su, Erdal Cayirci,and Yogesh sankarasubramaniam IEEE Communications Magazine 2002 Speaker:earl A Survey on Sensor Networks.
Energy-Efficient Signal Processing and Communication Algorithms for Scalable Distributed Fusion.
1 Collaborative Processing in Sensor Networks Lecture 2 - Mobile-agent-based Computing Hairong Qi, Associate Professor Electrical Engineering and Computer.
Collaborative Sampling in Wireless Sensor Networks Minglei Huang Yu Hen Hu 2010 IEEE Global Telecommunications Conference 1.
SIMPLE: Stable Increased Throughput Multi-hop Link Efficient Protocol For WBANs Qaisar Nadeem Department of Electrical Engineering Comsats Institute of.
Lecture 4 TTH 03:30AM-04:45PM Dr. Jianjun Hu CSCE569 Parallel Computing University of South Carolina Department of.
BARD / April BARD: Bayesian-Assisted Resource Discovery Fred Stann (USC/ISI) Joint Work With John Heidemann (USC/ISI) April 9, 2004.
Multiuser Receiver Aware Multicast in CDMA-based Multihop Wireless Ad-hoc Networks Parmesh Ramanathan Department of ECE University of Wisconsin-Madison.
Lecture 2: Statistical learning primer for biologists
The Problem of Location Determination and Tracking in Networked Systems Weikuan Yu, Hui Cao, and Vineet Mittal The Ohio State University.
Indexing Correlated Probabilistic Databases Bhargav Kanagal, Amol Deshpande University of Maryland, College Park, USA SIGMOD Presented.
Collaborative Broadcasting and Compression in Cluster-based Wireless Sensor Networks Anh Tuan Hoang and Mehul Motani National University of Singapore Wireless.
Multimedia Systems and Communication Research Multimedia Systems and Communication Research Department of Electrical and Computer Engineering Multimedia.
Energy-Efficient Signal Processing and Communication Algorithms for Scalable Distributed Fusion.
Bing Wang, Wei Wei, Hieu Dinh, Wei Zeng, Krishna R. Pattipati (Fellow IEEE) IEEE Transactions on Mobile Computing, March 2012.
Indian Institute of Technology Bombay 1 Communication Networks Prof. D. Manjunath
Markov Networks: Theory and Applications Ying Wu Electrical Engineering and Computer Science Northwestern University Evanston, IL 60208
On Mobile Sink Node for Target Tracking in Wireless Sensor Networks Thanh Hai Trinh and Hee Yong Youn Pervasive Computing and Communications Workshops(PerComW'07)
Distributed cooperation and coordination using the Max-Sum algorithm
Wireless Access and Networking Technology (WANT) Lab. An Efficient Data Aggregation Approach for Large Scale Wireless Sensor Networks Globecom 2010 Lutful.
Wireless sensor and actor networks: research challenges Ian. F. Akyildiz, Ismail H. Kasimoglu
Lecture 8: Wireless Sensor Networks By: Dr. Najla Al-Nabhan.
In the name of God.
Wireless Sensor Network Architectures
Introduction to Wireless Sensor Networks
StatSense In-Network Probabilistic Inference over Sensor Networks
المشرف د.يــــاســـــــــر فـــــــؤاد By: ahmed badrealldeen
Parallel Programming in C with MPI and OpenMP
Overview: Chapter 2 Localization and Tracking
Information Sciences and Systems Lab
Presentation transcript:

DISTIN: Distributed Inference and Optimization in WSNs A Message-Passing Perspective SCOM Team

Monitoring Data Aggregation Applications Sensor Selection (Clustering), Correlated Data Aggregation, Event Detection, State Estimation (Localization and Tracking, Activity Recognition). Distributed Inference? Issues: resource-constrained sensor nodes, in the presence of packet losses and link failures.

(1324,1245) Localization & Tracking Goal: Developing efficient, scalable, robust message- passing algorithms for distributed optimization & inference in large-scale sensor-actuator networks. Target detection Data fusion (if different sensors) Target localization Target classification (if multiple targets) Target tracking Transfer to sink & next leader Applications Sensor Selection (Clustering), Correlated Data Aggregation, Event Detection, State Estimation (Localization and Tracking, Activity Recognition). Distributed Inference?

Jointly optimizing inference, networking & comm. Hierarchical network architecture, cross-layer optimization Research Methodology Fusion center Decentralized Ad hoc Gossiping Hybrid Hierarchical Structure Hierarchical network is proved to be more scalable and energy-efficient

Research Methodology Leveraging probabilistic graphical models and message passing for representation & inference Problem Formulation (Global Maximization/Marginalization) Graphical Modeling (Factor Graph, MRF, etc) Message-Passing Rules (Min-Sum, Sum-Product, etc) Distributed Algorithms (Robust, Energy-Efficient, Scalable) Current heuristics are not efficient for WSN (slow convergence, not scale well, computation & comm. costs increase exponentially with network size)

Probabilistic Graphical Models & Message-Passing Algorithms Computing Graph meets Communications Graph – Capture the structure of a sensor network (for modeling statistical dependencies and/or communication links). – Parallel nature of message-passing operations (flexible message scheduling) Recently significant progress in PGM & MEPA – Junction Tree: exact solution, provable convergence, but exponential cost & not parallel – LoopyBP: approximate solutions, sufficient conditions of convergence, – Message-representation, message-censoring/damping. In-network processing & actuation – Each node of the network obtains the posterior distribution/optimal values of its variables. F. R. Kschischang et. al. “Factor Graph and the Sum-Product Algorithm”, IEEE Trans. Info. Theo. ‘01

Results: Min-Sum Clustering Algo Convergence rate Approximation Scalability Robustness Efficiency Min-Cost Hierarchical Architecture for Correlated Data Aggregation

Results: Communication Cost Average communication cost of MCDA is approximately ½ of MEGA’s MCDA 1 : trade-off between node residual energy and transmission cost with re-clustering after a constant number of rounds MCDA 2 : minimize transmission cost by exploiting data correlation without considering node residual energy (no re-clustering) MEGA: Minimum Energy Gathering Algorithm

Results: Network Lifetime Network lifetime is computed as the number of rounds until the first node dies. Network lifetime using MCDA is higher than MEGA because MCDA can balance between the transmission cost and node residual energy, resulted in lower re-clustering rate, which is an energy wasted process. MCDA 1 : trade-off between node residual energy and transmission cost with re-clustering after a constant number of rounds MCDA 2 : minimize transmission cost by exploiting data correlation without considering node residual energy (no re-clustering) MEGA: Minimum Energy Gathering Algorithm

Software Architecture for DISTIN MW Core Components: Message-Passing Inference Engine (pluggable): +Update outgoing messages using new sensor readings +Update marginal distribution using incoming messages Message-Censoring & Buffering +If message is not “new”=> do not send (censoring) +If message lost => update using old (buffered message) +Marshalling/unmarshalling compact message Networking layer +Neighbor Discovery & Maintenance +Message Broadcasting & Receiving APIs: Inputs: +Graphical Models & Priors +Scheduling Rules Outputs: +Marginal Distribution, MAP, Avg., etc.

Ubiquitous Computing LAB Challenges NP-hard global optimization tasks, performed on resource-constrained sensor nodes, in the presence of packet losses and link failures. Current heuristics are not efficient for WSN (slow convergence, not scale well, computation & comm. costs increase exponentially with network size) Methodology Jointly optimizing inference, networking & comm. Leveraging probabilistic graphical models and message passing for representation & inference Contributions MEPA: Robust message-passing algorithms for efficient sensor clustering & correlated data aggregation: simple, fast, good approximation & highly localized Robust algorithms for planning & learning in collaborative multi-agent settings (e.g. state estimation, activity recognition, localization and tracking in WSN) - ongoing Broader Impacts WSN : MEPA as a macro-programming language for novel applications (structural health monitoring, precision agriculture, etc.) Graphical Models : novel mess. passing algorithms under comm. constraints (message representation, censoring, and scheduling) Advance the theory & practice in these fields Project UCLab, KHU Applications Sensor Selection (Clustering), Correlated Data Aggregation, Event Detection, State Estimation (Localization and Tracking, Activity Recognition). Research Focus Developing efficient, scalable, robust message- passing algorithms for distributed optimization & inference in large-scale sensor-actuator networks.