Home

WiSE-MNet

1. Gaussian Mixture Models For more details the reader can refer to Distributed target tracking under realistic network conditions in the proceeding of Sensor Signal Processing for Defence SSPD 2011 London UK 27 29 September 2011 3 Installation WiSE MNet is based on OMNeT and is an extension of the Castalia simulation model WiSE MNet has been developed using the version 4 1 of OMNeT and the 3 1 version of Castalia Although OMNeT is available for Windows systems Castalia has been designed for GNU Linux like systems see Castalia reference manual For this reason we strongly recommend to used a GNU Linux like system to use WiSE MNet the Ubuntu GNU Linux distribution has been successfully used However installation for Windows systems might be possible through the Cygwin environment although this has not been tested Prerequisite e OMNeT 4 1 e OpenCV 3 1 Prerequisite Installing OMNeT For full instructions and details about the installation of OMNeT refer to the Linux section of the OMNeT installation guide The following steps should be performed for a fresh installation of OMNeT We assume to work in the home directory type cd to enter it in a bash shell 1 Get the OMNeT 4 1 sources from the download page 2 Extract the source files tar xvzf omnetpp 4 1 src tgz A folder omnetpp 4 1 will be created 3 Set the environment variables to point to the OMNeT binar
2. Y y Y 4 Wireless Channel to from Wireless Channel to Wireless Channel Figure 2 Generic data types WiseBaseApplication WiseBaseApplication WiseBaseApplication IS Diea Communication R WiseBaseApplication ry 4 LA Communication Module Communication Module A 4 i Li WiseDummyWirelessChannel Wireless Channel a b Figure 3 Idealistic communication through a WiseDummyWirelessChannel and b DirectApplicationMes sage The WiseDummyWirelessChannel is a module that is used to bypass the Castalia communication stack and wireless channel This module allows to specify the node neighborhood and performs idealistic commu nication with no delay or packet loss corruption The module is to be used alternatively to the Wireless Channel module proposed in Castalia The interaction between application and communication module does not change the user can decide whether to change the network capability in the simulation configuration without changing the application logic The second mechanism provided for idealistic communication is the DirectApplicationMessage This is an OMNeT communication port that has been added to the application module so that two nodes application layers can interact directly without bypassing the communication modules With this mecha nismm a part of the application can use a realistic or idealistic network communication through either the WirelessChannel or the WiseDummyWirelessChannel while
3. and outside the FOVs e wise Simulations WiseCameraDPF_example3 omnetpp ini 20 cameras and a single moving target This simulation folders contain a Makefile to run and clean the simulation output To run the simulation enter the setup directory and type make Several files will be created after the simulation with GUI animation The files dpf_results txt and dpf_part_results txt contain information respectively about the tracking output and the intermediate tracking steps To clean up the simulation folder type make clean 10
4. application We assume the reader to be familiar with the OMNeT environment and to know the basics about the Castalia simulation model The reference versions over OMNeT and Castalia are respectively the 4 1 and the 3 1 Documentations and tutorials about OMNeT can be found at the project documentation page http www omnetpp org documentation We particularly suggest to read the first and then to use the API Reference when developing new modules A copy of the Castalia user s manual is available at http castalia npc nicta com au documentation php 2 Overview In this section we present an overview of WiSE MNet The overall structure of the network and node model is depicted in Figure l to from Physical Process WiseBasePhysicalProcess 4 Y WiseBaseSensorManager Resource T Manager po 1 WiseBasePhysicalProcess to from other lt 4 WiseBaseApplication E 4 4 WiseBaseApplication T Y y d Node Wise aa Node Wise ace Node Wise 1 Mobility 1 T I Communication Module Manager Ld Y Y A i j Wireless Channel Dummy Channel to from Wireless Channel to Wireless Channel a Network Model b Node Wise Model Figure 1 WiSE MNet network and node model overview 2 1 Generalized sensor data types The generalization of the sensor data types is obtained by defining an abstract class Wis
5. E MNet properly installed 1 Enter the simulation directory cd WiSE MNet v0 1 wise Simulations WiseTracker_test 2 Run the simulation using WiSE MNet extended Castalia Castalia c General Running configuration 1 1 A WORLD window will pop up showing a simple 2D representation of the ground plane world the cameras and the targets 3 Press a button on the WORLD window to start the simulation This will show an animation of two targets moving with different types of motion on the ground plane with 4 camera nodes 3 with partially overlapping FOV 4 When the simulation is over a myLog txt will be created containing the application printouts 4 4 WiseCameraDPF This module implements a Distributed Particle Filter DPF tracker based on a sequential aggregation mechanism to exchange the Partial Posterior see Section B Source files WiseCameraDPF ned WiseCameraDPF h WiseCameraDPF cc WiseCameraDPFMessage msg WiseCameraDPFMessage_custom h Run the example s Three different simulation setups have been provided to test this algorithm e wise Simulations WiseCameraDPF_example1 omnetpp ini shows an example with 4 camera nodes with partially overlapping FOV and two targets moving inside a fully overlapping region all the cameras observing the target e wise Simulations WiseCameraDPF_example2 omnetpp ini 4 cameras with partially and non overlapping FOV and a single target moving inside
6. SE MNet makefiles makeclean Script to properly clean up the WiSE MNet build others The WiSE MNet simulation setup files are contained in the wise Simulations sub folder The defini tions redefinitions of the OMNeT modules NED C files can be found in the wise src sub folder and in particular the wise src wise contains the main part of the software The structure of the wise src wise subtree is the following wise src wise node Definition of the node s components world Definition of the world s elements PhysicalProcess terrain wirelessChannel WirelessChannel and WiseDummyWirelssChannel gui simple GUI code utils Utilities GMM ParticleFilter helper classes 2 NOTE if you have a multi core machine compilation will be faster by running the make command with the j option and passing the number of cores plus one as argument For instance in a dual core machine use make j 3 The structure of the wise src wise node subtree is the following node sensorManager Sensor Manager modules wiseEmptySensorManager Dummy sensor producing random numbers wiseCameraManager Camera Manager module node application Application modules wiseCameraApplication WiseCameraApplication base class wiseCameraSimplePeriodicTracker WiseCameraSimplePeriodicTracker base class wiseAppTest Example WiseAppTest module wiseCameraAppTest Example WiseCameraAppTest module wiseCameraTrackerTest Example WiseCameraTrackerTest m
7. WiSE MNet Wireless Simulation Environment for Multimedia Networks User s Manual Christian Nastasi c nastasi sssup it nastasichr gmail com December 4 2011 Contents 2 1 Generalized sensor data tupes 2 2 dealistic communication mechanism 2 3 Simple GU 2 41 WiseMovinglarget 2 42 WiseCameraManager 3 1__Prerequisite 2 aaa 3 2__ Installing WiSE MNet 3 3 Software organization 4 Application Example 4 3 WiseCamera Tracker Test 4 4 _WiseCameraDP 1 Introduction The Wireless Simulation Environment for Multimedia Sensor Networks WiSE MNet has been designed to simulate distributed algorithms for Wireless Multimedia Sensor Networks WMSNs under realistic network conditions The simulation environment is based on one of the most popular network simulator Among the several simulation models for the OMNeT environment Castalia s the one that has been designed with similar goals although it focuses on classic Wireless Sensor Networks WSNs WiSE MNet is proposed as an extension of the Castalia OMNeT simulator The main extensions provided to the Castalia simulation model can be summarized as e generalization of the sensor data type from scalar based to any type e idealistic communication and direct application communication e simple GUI for 2D world representation e concrete modules for moving target camera modelling target tracking
8. ePhysicalPro cessMessage that has to be derived to define any type of physical process information Accordingly other abstract classes have been modified to redefine some of the original Castalia modules In particular e WiseBasePhysicalProcess WiseBaseSensorManager and WiseBaseApplication that redefine respec tively the base classes for the physical process the sensor manager and the application layer e WisePhysicalProcessMessage WiseSensorManagerMessage and WiseApplicationPacket information exchange classes 2 2 Idealistic communication mechanisms There are two idealistic communication mechanisms that have been introduced the WiseDummy Wireless Channel and the DirectApplicationMessage The first one changes the network properties to idealistic seemingly from the application point of view the second one is rather a magic direct information exchange channel Figure B represents the two mechanisms to from Physical Process WiseBasePhysicalProcess Li WiseBaseSensorManager H Resource Manager a 1 WiseBasePhysicalProcess WiseBaseApplication i ry ry ry C ji i D Node Wise Node Wise eee Node Wise Mobility I I I Communication Module Manager
9. is a simplified projection model that assumes a top down facing camera observing targets modelled according to the WiseMoving Target module 2 4 3 Application Layer classes The application module contains the algorithm of the distributed application The user should typically provide its own application module to implement a new distributed algorithm The application module de WiseBasePhysicalProcess WisePhysicalProcessMessage WiseBaseSensorManager WiseSensorManagerMessage Communication Module f i Wireless Channel Figure 5 WiSE MNet concrete modules rived from WiseBaseApplication interacts with a WiseBaseSensorManager and the Castalia communication module in order to realize the logic of the distributed algorithm In the current distribution of WiSE MNet we provided some application layer classes according to the hierarchy shown in Figure 6 WiseBaseApplication A WiseCameraManager WiseCameraApplication A WiseCameraSimplePeriodicTracker WiseCameraDPF Figure 6 Application class hierarchy WiseBaseApplication This is the base class for any application module in WiSE MNet The class provides a mechanism to automatically calculate the set of radio neighbor nodes a set of nodes that can be reached by a give node with a single hop wireless co
10. mmunication The application WiseApp Test shows an example of a class derived directly from this class WiseCameraApplication This is a base class to derive from when we are interested in creating an ap plication module that uses a WiseCameraManager During the startup phase this class will query the WiseCameraManager to collect information about the camera e g FOV and to create a list of overlapping FOV neighbor nodes The node s and other nodes camera information are available as protected member respectively called camera_info and overlapping _fov_cameras The application WiseCameraAppTest shows an example of a class derived from WiseCameraApplication WiseCameraSimplePeriodicTracker This class is derived from WiseCameraApplication and is meant to be used as a base class for periodic tracking algorithms based on WiseCameraApplication This class defines a set of callback like functions that will be called at different steps of any periodic tracking algorithm The class defines some functions that will be called at startup for initializations and other functions that will be periodically called when a new image is available The application WiseCameraTrackerTest is a basic example derived from WiseCameraSimplePeriodic Tracker WiseCameraDPF It is a WiseCameraSimplePeriodicTracker that implements a distributed particle filter algorithm The algorithm uses a sequential aggregation mechanism exchanging the partial posterior ap proximated with
11. n setup wise Simulations WiseCamera_test omentpp ini should be used We assume to work in the home directory type cd to enter it in a bash shell and to have WiSE MNet properly installed 1 Enter the simulation directory cd WiSE MNet v0 1 wise Simulations WiseCamera_test 2 Run the simulation using WiSE MNet extended Castalia Castalia c General Running configuration 1 1 A WORLD window will pop up showing a simple 2D representation of the ground plane world the cameras and the targets 3 Press a button on the WORLD window to start the simulation This will show an animation of the three targets moving with different types of motion on the ground plane 4 When the simulation is over a myLog txt will be created containing the application printouts 4 3 WiseCameraTrackerTest The module is derived from the WiseCameraSimplePeriodicIracker class This example shows how the skeleton of a distributed target tracking application looks like when using the WiseCameraManager pro ducing target detections and adopting a classic periodic tracker approach Source files WiseCameraTrackerTest ned WiseCameraTrackerTest h WiseCameraTrackerTest cc WiseCameraTrackerTestMessage msg Run the example To try this example the simulation setup wise Simulations WiseTracker_test omentpp ini should be used We assume to work in the home directory type cd to enter it in a bash shell and to have WiS
12. odule wiseCameraDPF WiseCameraDPF tracker 4 Application Examples 4 1 WiseApp Test The module is derived directly from the WiseBaseApplication base class This is the simplest application example and shows how to use the three basic elements of any distributed application for WMSNs sensor reading network communication and time triggered actions Source files WiseAppTest ned WiseAppTest h WiseAppTest cc Run the example To try this example the simulation setup wise Simulations WiseSimpleApp test omentpp ini should be used We assume to work in the home directory type cd to enter it in a bash shell and to have WiSE MNet properly installed 1 Enter the simulation directory cd WiSE MNet v0 1 wise Simulations WiseSimpleApp_test 2 Run the simulation using WiSE MNet extended Castalia Castalia c General Running configuration 1 1 A myLog txt will be created containing the application printouts 4 2 WiseCameraApp Test The module is derived from the WiseCameraApplication class This example module is similar to the WiseApp Test one It shows a basic interaction with the WiseCameraManager a camera based sensor man ager that produces target detections The example shows also a custom application message exchanged among the nodes Source files WiseCameraAppTest ned WiseCameraAppTest h WiseCameraAppTest cc WiseCameraAppTestPacket msg Run the example To try this example the simulatio
13. some part might assume ideal node to node interaction 2 3 Simple GUI We included a simple GUI that can be useful for testing and evaluation of distributed algorithm for WMSNs We currently used the GUI for a simple representation of a 2D world ground plane where targets and sensor cameras can be displayed during the simulation see Figure 4 The GUI could be further used to evaluate distributed algorithms involving computer vision processing 2 4 Concrete modules In the current distribution of WiSE MNet we included some concrete classes that have been used to simulate distributed target tracking algorithms in simplified context target camera center camera FOV Figure 4 Simple GUI for 2D world scenarios 2 4 1 WiseMoving Target This module is a WiseBasePhysicalProcess that implements moving target in a 2D ground plane Targets are currently represented as bounding boxes and can move according to different types of motion linear circular linear circular and random 2 4 2 WiseCameraManager This module is a WiseBaseSensorManager that implements the sensing logic of the node s camera The module is strongly related to the type of physical process we are using The WiseCameraManager has been designed to support different types of sensing through the WiseCameraHandler mechanism allowing the user to easily add different camera models e g projection models We currently support only the WiseCameraDetections model which
14. talia We assume to work in the home directory type cd to enter it in a bash shell 1 Extract the source files tar xvzf WiSE MNet v0 1 tar gz cd WiSE MNet v0 1 A folder WiSE MNet v0 1 will be created 2 Set the environment variables to point to the OMNeT binary paths export PATH PATH WiSE MNet v0 1 bin This line should be also appended to the bashrc file 3 Create the makefiles to compile Castalia with the WiSE MNet extensions makemake 4 Build make 5 To properly clean the last Castalia build the following can be used makeclean 3 3 Software organization The WiSE MNet root directory according to installation instruction is WiSE MNet v0 1 contains the original Castalia source files and the extensions provided by WiSE MNet All WiSE MNet files used to define redefine modules and to run simulations are contained in the wise folder in the root directory The native Castalia files can still be found in their original position src and Simulations In this sec tion we give an overview of the software organization in folders to help the reader browsing the source code The structure of the WiSE MNet root directory is bin native Castalia python scripts src native Castalia NED C sources Simulation native Castalia simulation setups wise WiSE MNet NED C sources and Simulation setups makemake Script to configure the Wi
15. y paths export PATH PATH omnetpp 4 1 bin export LD_LIBRARY_PATH omnetpp 4 1 1lib These two lines should be also appended to the bashrc file 4 Compile OMNeT cd omnetpp 4 1 configure make 5 OMNeT should be successfully installed The following command can be used to verify that the OMNeT executables are in the execution path which opp_makmake Installing OpenCV For installation instruction of the OpenCV library please refer to http opencv willowgarage com wiki InstallGuide A binary of the OpenCV library is available for recent Ubuntu distributions In such case the following command could be used for installation sudo apt get install libcv dev libcvaux dev libhighgui dev To check whether the OpenCV library are correctly installed pkg config opencv cflags pkg config opencv libs These should print the include paths compiler and linker options required to build the WiSE MNet with the OpenCV library 1 NOTE if you have a multi core machine compilation will be faster by running the make command with the j option and passing the number of cores plus one as argument For instance in a dual core machine use make j 3 3 2 Installing WiSE MNet The simulator source files are distributed as an extension of the Castalia simulation model The steps required to compile the simulator are equivalent to those for Cas

WiSE-MNet

Contents

Download Pdf Manuals

Related Search

Related Contents