Self-organising zooms for decentralised redundancy management in visual sensor networks

Lukas Esterle; Bernhard Rinner; Peter Lewis

doi:10.1109/SASO.2015.12

Self-organising zooms for decentralised redundancy management in visual sensor networks

Lukas Esterle, Bernhard Rinner, Peter Lewis

Computer Science Research Group

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

When visual sensor networks are composed of cameras which can adjust the zoom factor of their own lens, one must determine the optimal zoom levels for the cameras, for a given task. This gives rise to an important trade-off between the overlap of the different cameras’ fields of view, providing redundancy, and image quality. In an object tracking task, having multiple cameras observe the same area allows for quicker recovery, when a camera fails. In contrast having narrow zooms allow for a higher pixel count on regions of interest, leading to increased tracking confidence. In this paper we propose an approach for the self-organisation of redundancy in a distributed visual sensor network, based on decentralised multi-objective online learning using only local information to approximate the global state. We explore the impact of different zoom levels on these trade-offs, when tasking omnidirectional cameras, having perfect 360-degree view, with keeping track of a varying number of moving objects. We further show how employing decentralised reinforcement learning enables zoom configurations to be achieved dynamically at runtime according to an operator’s preference for maximising either the proportion of objects tracked, confidence associated with tracking, or redundancy in expectation of camera failure. We show that explicitly taking account of the level of overlap, even based only on local knowledge, improves resilience when cameras fail. Our results illustrate the trade-off between maintaining high confidence and object coverage, and maintaining redundancy, in anticipation of future failure. Our approach provides a fully tunable decentralised method for the self-organisation of redundancy in a changing environment, according to an operator’s preferences.

Original language	English
Title of host publication	Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015
Publisher	IEEE
Pages	41-50
Number of pages	10
ISBN (Print)	978-1-4673-7535-1
DOIs	https://doi.org/10.1109/SASO.2015.12
Publication status	Published - 29 Oct 2015
Event	9th IEEE International Conference on Self-Adaptive and Self-Organizing Systems - Cambridge, MA, United States Duration: 21 Sept 2015 → 25 Sept 2015

Publication series

Name	2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems
Publisher	IEEE
ISSN (Print)	1949-3673
ISSN (Electronic)	1949-3681

Conference

Conference	9th IEEE International Conference on Self-Adaptive and Self-Organizing Systems
Country/Territory	United States
City	Cambridge, MA
Period	21/09/15 → 25/09/15

Bibliographical note

© 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Funding: Austrian Institute of Technology and the Austrian Federal
Ministry of Science, Research and Economy HRSM program.

Keywords

visual sensor networks
edundancy management
decentralised learning
self-organisation
runtime trade-offs

Access to Document

10.1109/SASO.2015.12

Self-organising zooms for decentralised redundancy management in visual sensor networks
© 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Accepted author manuscript, 935 KB

Cite this

Esterle, L., Rinner, B., & Lewis, P. (2015). Self-organising zooms for decentralised redundancy management in visual sensor networks. In Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015 (pp. 41-50). (2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems). IEEE. https://doi.org/10.1109/SASO.2015.12

@inproceedings{5a2a2359f4fb4a0ab29367a547dde89c,

title = "Self-organising zooms for decentralised redundancy management in visual sensor networks",

abstract = "When visual sensor networks are composed of cameras which can adjust the zoom factor of their own lens, one must determine the optimal zoom levels for the cameras, for a given task. This gives rise to an important trade-off between the overlap of the different cameras{\textquoteright} fields of view, providing redundancy, and image quality. In an object tracking task, having multiple cameras observe the same area allows for quicker recovery, when a camera fails. In contrast having narrow zooms allow for a higher pixel count on regions of interest, leading to increased tracking confidence. In this paper we propose an approach for the self-organisation of redundancy in a distributed visual sensor network, based on decentralised multi-objective online learning using only local information to approximate the global state. We explore the impact of different zoom levels on these trade-offs, when tasking omnidirectional cameras, having perfect 360-degree view, with keeping track of a varying number of moving objects. We further show how employing decentralised reinforcement learning enables zoom configurations to be achieved dynamically at runtime according to an operator{\textquoteright}s preference for maximising either the proportion of objects tracked, confidence associated with tracking, or redundancy in expectation of camera failure. We show that explicitly taking account of the level of overlap, even based only on local knowledge, improves resilience when cameras fail. Our results illustrate the trade-off between maintaining high confidence and object coverage, and maintaining redundancy, in anticipation of future failure. Our approach provides a fully tunable decentralised method for the self-organisation of redundancy in a changing environment, according to an operator{\textquoteright}s preferences.",

keywords = "visual sensor networks, edundancy management, decentralised learning, self-organisation, runtime trade-offs",

author = "Lukas Esterle and Bernhard Rinner and Peter Lewis",

note = "{\textcopyright} 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Funding: Austrian Institute of Technology and the Austrian Federal Ministry of Science, Research and Economy HRSM program.; 9th IEEE International Conference on Self-Adaptive and Self-Organizing Systems ; Conference date: 21-09-2015 Through 25-09-2015",

year = "2015",

month = oct,

day = "29",

doi = "10.1109/SASO.2015.12",

language = "English",

isbn = "978-1-4673-7535-1",

series = "2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems",

publisher = "IEEE",

pages = "41--50",

booktitle = "Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015",

address = "United States",

}

Esterle, L, Rinner, B & Lewis, P 2015, Self-organising zooms for decentralised redundancy management in visual sensor networks. in Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015. 2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems, IEEE, pp. 41-50, 9th IEEE International Conference on Self-Adaptive and Self-Organizing Systems, Cambridge, MA, United States, 21/09/15. https://doi.org/10.1109/SASO.2015.12

Self-organising zooms for decentralised redundancy management in visual sensor networks. / Esterle, Lukas; Rinner, Bernhard; Lewis, Peter.
Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015. IEEE, 2015. p. 41-50 (2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems).

Research output: Chapter in Book/Published conference output › Conference publication

TY - GEN

T1 - Self-organising zooms for decentralised redundancy management in visual sensor networks

AU - Esterle, Lukas

AU - Rinner, Bernhard

AU - Lewis, Peter

N1 - © 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. Funding: Austrian Institute of Technology and the Austrian Federal Ministry of Science, Research and Economy HRSM program.

PY - 2015/10/29

Y1 - 2015/10/29

N2 - When visual sensor networks are composed of cameras which can adjust the zoom factor of their own lens, one must determine the optimal zoom levels for the cameras, for a given task. This gives rise to an important trade-off between the overlap of the different cameras’ fields of view, providing redundancy, and image quality. In an object tracking task, having multiple cameras observe the same area allows for quicker recovery, when a camera fails. In contrast having narrow zooms allow for a higher pixel count on regions of interest, leading to increased tracking confidence. In this paper we propose an approach for the self-organisation of redundancy in a distributed visual sensor network, based on decentralised multi-objective online learning using only local information to approximate the global state. We explore the impact of different zoom levels on these trade-offs, when tasking omnidirectional cameras, having perfect 360-degree view, with keeping track of a varying number of moving objects. We further show how employing decentralised reinforcement learning enables zoom configurations to be achieved dynamically at runtime according to an operator’s preference for maximising either the proportion of objects tracked, confidence associated with tracking, or redundancy in expectation of camera failure. We show that explicitly taking account of the level of overlap, even based only on local knowledge, improves resilience when cameras fail. Our results illustrate the trade-off between maintaining high confidence and object coverage, and maintaining redundancy, in anticipation of future failure. Our approach provides a fully tunable decentralised method for the self-organisation of redundancy in a changing environment, according to an operator’s preferences.

AB - When visual sensor networks are composed of cameras which can adjust the zoom factor of their own lens, one must determine the optimal zoom levels for the cameras, for a given task. This gives rise to an important trade-off between the overlap of the different cameras’ fields of view, providing redundancy, and image quality. In an object tracking task, having multiple cameras observe the same area allows for quicker recovery, when a camera fails. In contrast having narrow zooms allow for a higher pixel count on regions of interest, leading to increased tracking confidence. In this paper we propose an approach for the self-organisation of redundancy in a distributed visual sensor network, based on decentralised multi-objective online learning using only local information to approximate the global state. We explore the impact of different zoom levels on these trade-offs, when tasking omnidirectional cameras, having perfect 360-degree view, with keeping track of a varying number of moving objects. We further show how employing decentralised reinforcement learning enables zoom configurations to be achieved dynamically at runtime according to an operator’s preference for maximising either the proportion of objects tracked, confidence associated with tracking, or redundancy in expectation of camera failure. We show that explicitly taking account of the level of overlap, even based only on local knowledge, improves resilience when cameras fail. Our results illustrate the trade-off between maintaining high confidence and object coverage, and maintaining redundancy, in anticipation of future failure. Our approach provides a fully tunable decentralised method for the self-organisation of redundancy in a changing environment, according to an operator’s preferences.

KW - visual sensor networks

KW - edundancy management

KW - decentralised learning

KW - self-organisation

KW - runtime trade-offs

UR - http://www.scopus.com/inward/record.url?scp=84959262315&partnerID=8YFLogxK

U2 - 10.1109/SASO.2015.12

DO - 10.1109/SASO.2015.12

M3 - Conference publication

AN - SCOPUS:84959262315

SN - 978-1-4673-7535-1

T3 - 2015 IEEE 9th International Conference on Self-Adaptive and Self-Organizing Systems

SP - 41

EP - 50

BT - Proceedings : 2015 IEEE Ninth International Conference on Self-Adaptive and Self-Organizing Systems, SASO 2015

PB - IEEE

T2 - 9th IEEE International Conference on Self-Adaptive and Self-Organizing Systems

Y2 - 21 September 2015 through 25 September 2015

ER -

Self-organising zooms for decentralised redundancy management in visual sensor networks

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this