WiSE-MNet: an experimental environmentfor wireless multimedia sensor networks   Christian Nastasi                     Andr...
WMSNs: context • Wireless Multimedia Sensor Networks (WMSNs)        – Evolution of WSNs to more complex (vectorial) data -...
Network simulators • Several existing network simulators        –   NS-2        –   OPNET        –   TrueTime        –   O...
NS-2: a popular open-source network simulator • Modules written in C++ • Simulation interface based on O-Tcl • Originally ...
OPNET: a commercial network simulation software   .   • Modules written in C++   • GUI for simulation     scenarios and ne...
TrueTime: a Matlab/Simulink toolbox  • Mainly for control applications         – to simulate (distributed) control  • Node...
OMNeT++: open-source (generic) network simulator• Generic discrete-event simulation engine• Generic modules interactions c...
Which simulation model? • Simulation Model: Castalia        –   Designed to model distributed algorithm for WSNs        – ...
Castalia simulation model                                          Physical Process                                       ...
Proposed extensions to the Castalia model  • Generalization of the sensor data-type        – from scalar-based to any data...
Generalized data-types  Castalia supports only scalar-based applications   Scalar function                                ...
Generalized data-types  We want to have any possible data-type: object-oriented approach      Any process                 ...
Idealistic communication  Seemingly from the application point of view: bypass communication                              ...
New specific modules • WiseMovingTarget - physical process        – moving target on a ground plane        – different typ...
Simulation example  Tracking algorithm        –   Distributed Particle Filter (DPF)        –   Sequential measurement aggr...
Network delay problem     Communication affects tracking performance                                               f(x k |...
Simulation setup   • Network          –   T-MAC protocol, BW = 250 kbps          –   request-to-send/clear-to-send mechani...
Performance degradation due to network delayC. Nastasi, A. Cavallaro   WiSE-MNet: an experimental environment for wireless...
Conclusions • Conclusions        – Realistic simulation environment for WMSNs               • Extension of the Castalia/OM...
Upcoming SlideShare
Loading in …5
×

WiSE-MNet: an experimental environment for Wireless Multimedia Sensor Networks

1,648 views

Published on

Published in: Technology, Business
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,648
On SlideShare
0
From Embeds
0
Number of Embeds
2
Actions
Shares
0
Downloads
0
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide

WiSE-MNet: an experimental environment for Wireless Multimedia Sensor Networks

  1. 1. WiSE-MNet: an experimental environmentfor wireless multimedia sensor networks Christian Nastasi Andrea Cavallaro c.nastasi@sssup.it andrea.cavallaro@eecs.qmul.ac.uk www.eecs.qmul.ac.uk/~andrea/wise-mnet.html
  2. 2. WMSNs: context • Wireless Multimedia Sensor Networks (WMSNs) – Evolution of WSNs to more complex (vectorial) data - e.g. audio, video – Several challenges • high network scalability • resource-constrained devices • raw data cannot be (generally) transferred • local processing is required (but it is much more complex than in WSNs) – Multi-disciplinary problem • networking • sensor signal processing • embedded system design Need a simulation environment for WMSNs!C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  3. 3. Network simulators • Several existing network simulators – NS-2 – OPNET – TrueTime – OMNeT++ • What are we looking for? – Configurability (e.g. scripting languages, configuration interfaces) – Support for standard and state-of-the-art communication protocols – Extensibility – Programming languageC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  4. 4. NS-2: a popular open-source network simulator • Modules written in C++ • Simulation interface based on O-Tcl • Originally designed for TCP/IP computer networks – Extensions to support wireless network (external contribution) – Integration of wireless and wired networks not simple – Mobility requires extensions • Network stacks – generally much more complex than WSN’s stacks • Maintainability and extensibility are difficult http://www.isi.edu/nsnam/ns/C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  5. 5. OPNET: a commercial network simulation software . • Modules written in C++ • GUI for simulation scenarios and network models • Three levels of configurability – Network level • defines topology – Node level • defines node behaviour and data flow – Process level • defines protocols’ logic through finite state machine formalism http://www.opnet.com/C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  6. 6. TrueTime: a Matlab/Simulink toolbox • Mainly for control applications – to simulate (distributed) control • Node behaviour in Matlab/Simulink • Configuration through Simulink • Simple networking models – including IEEE 802.15.4 • Integration with Matlab-based software is straightforward http://www3.control.lth.se/truetime/C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  7. 7. OMNeT++: open-source (generic) network simulator• Generic discrete-event simulation engine• Generic modules interactions can be defined – behaviour is coded in C++ – interconnections/composition specified through a Network Description (NED) language – parameters can be set through configuration files• Highly flexible and extensible with external libraries• Network elements – nodes, protocols, channels – provided (externally) as simulation models (INET, MiXiM, Castalia) http://www.omnetpp.org/C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  8. 8. Which simulation model? • Simulation Model: Castalia – Designed to model distributed algorithm for WSNs – Realistic wireless channel and radio models – Physical process (real-world events) and sensing model – Node Mobility – Resource management (e.g. CPU, battery) http://castalia.npc.nicta.com.au/C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  9. 9. Castalia simulation model Physical Process Physical Process Physical Process Node Node ... Node Wireless Channel to/from Physical Process Sensor Manager Resource Manager Application Mobility Communication Manager to/from Wireless Channel to/from Wireless ChannelC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  10. 10. Proposed extensions to the Castalia model • Generalization of the sensor data-type – from scalar-based to any data type • Idealistic communication & direct application communication • Specific modules – moving target – simple camera modelling – target tracking application • Simple GUI – for 2D world representationC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  11. 11. Generalized data-types Castalia supports only scalar-based applications Scalar function Physical Process Sensor Manager Scalar-based Scalar value Application ... Communication Wireless ChannelC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  12. 12. Generalized data-types We want to have any possible data-type: object-oriented approach Any process Physical Process WiseBasePhysicalProcess WiseBasePhysicalProcessMessage Sensor Manager WiseBaseSensorManager Any data-type WiseBaseSensorManagerMessage Application WiseBaseApplication ... WiseBaseApplicationPacket Communication Wireless ChannelC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  13. 13. Idealistic communication Seemingly from the application point of view: bypass communication Physical Process Sensor Manager Sensor Manager Application Application Communication Communication Wireless Channel WiseDummyWirelessChannelC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  14. 14. New specific modules • WiseMovingTarget - physical process – moving target on a ground plane – different types of motion: e.g. linear, circular, random – target represented as a box • WiseCameraManager - sensor manager – top-down facing camera – object detections produced: bounding boxes – can be extended to any FOV, projection model or to use real-world images • WiseCameraApplication - application layer – hierarchy of modules implementing a distributed tracking algorithmC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  15. 15. Simulation example Tracking algorithm – Distributed Particle Filter (DPF) – Sequential measurement aggregation through partial posterior exchange – Guassian Mixture Modelling (GMM) to approximate the partial posterior – # Particles = 500 1:4 f(x k | Z1:k 1 , z k )  f(x k | Z1:k ) 1 f(x k | Z1:k 1 , z k ) start estimate 4 zk 1 zk 2 zk f(x k | Z1:k 1 , z k:2 ) 1 3 zk f(x k | Z1:k 1 , z k:3 ) 1Distributed target tracking under realistic network conditionsC. Nastasi, A. CavallaroProc. of Sensor Signal Processing for Defence, London, 28-29 September 2011C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  16. 16. Network delay problem Communication affects tracking performance f(x k | Z1:k 1 , z k:4 )  f(x k | Z1:k ) 1 1 f(x k | Z1:k 1 , z k ) start estimate 4 zk 1 zk ∆t ∆t 2 zk f(x k | Z1:k 1 , z k:2 ) 1 3 zk f(x k | Z1:k 1 , z k:3 ) 1 ∆t Analysis of performance degradation with increased network load • 0-GMM (exchanging the whole particle set) • 5-GMM (mixture of 5 Gaussians to approximate the particle set)C. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  17. 17. Simulation setup • Network – T-MAC protocol, BW = 250 kbps – request-to-send/clear-to-send mechanism – acknowledged-transmission mechanism – number of retransmissions: 10 • Cameras – Covering 6000 sqm – 500 sensors (random uniform distribution) – 6m from the ground plane (top-down facing) – Frame Rate = 1fpsC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  18. 18. Performance degradation due to network delayC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks
  19. 19. Conclusions • Conclusions – Realistic simulation environment for WMSNs • Extension of the Castalia/OMNeT++ simulation model • Generic data-type support • Simplistic world model – Application example: distributed target tracking Simulator available as open source at www.eecs.qmul.ac.uk/~andrea/wise-mnet.html • Future work – More complex networking conditions – Support for real-world imagesC. Nastasi, A. Cavallaro WiSE-MNet: an experimental environment for wireless multimedia sensor networks

×