Distributed adaptive MAC protocol for WSN-based wildlife protection

Get Complete Project Material File(s) Now! »

Particular WSN application usecase: LIRIMA PREDNET project

The Inria FUN3 team of Inria Lille is collaborating with Stellenbosch Uni-versity in South Africa in the framework of the project LIRIMA PREDNET4. The aim of the project is to develop a Wireless Sensor Network (WSN) that is able to operate in sparsely populated outlying rural and wilderness areas, for efficient monitoring and protection of resources and ecosystems. One case of application of the PREDNET WSN is related to the rhinoceros track-ing and protection. These animals become more and more often victims of poaching because of their horns [27, 95]. In the present moment about 9000 rhinoceros live in the southern part of the Kruger Park (the habitat area is about 5000 square km ) – the potential zone of implementation of the system. Thus, the current average density of rhinoceros in Kruger Park is about 1.7 animals per square kilometer. The WSN should help final user (biologists, zoologists) to regularly gather information about every animal of targeted specie (e.g. position, heart rate, acceleration data…) within the zone of monitoring (normal mode). Moreover, the WSN has to deliver the urgent alarm messages from animals in danger (alarm mode).

Functional requirements

South African specialists have established the following key requirements for the animal tracking system in order to fit the application needs and make it efficient: The data provided by different sensors (GPS, pulse oxymeter, ac-celerometer …) must be sent at least every 15 minutes. Data are gathered at different rates (depending on the sensor type) between transmissions.
The WSN must be operational in the entire southern part of the Kruger Park where most of the rhino population lives. Weight of wearable device should not exceed 500 grams. Battery life must be guaranteed during several years. The robustness of communications must be guaranteed (especially for the alarm mode). While latency for the normal mode is not critical, the delivery time for an urgent alarm message must not exceed 1 minute.

Other wildlife tracking projects

Other wildlife tracking projects are described in literature. Even though the final aim of these projects is similar to the PREDNET project, there are several differences in the concept. For example, ZebraNet project[70] aims to track wild animals in central Kenya. Nevertheless the project focuses mainly on zebras, the developed network could be applied for different kind of animals. The authors em-phasis the fully mobile nature of the sensor network developed within the project. Not only end nodes (devices, carrying by zebras), but also base sta-tions (the data sinks) are considered mobile. The challenge of this solution is the unknown time of the BTS availability. It is supposed that the base sta-tion can arrive close to the animals between noon and midnight. The nodes, thus have to search for a BTS during long time which is energy consuming. Moreover, that introduces huge delays in data collection (at least 12 hours). These factors make this solution unsuitable to meet the requirements of the PREDNET project (see the Section 1.3.1). This fact also makes ZebraNet so-lution unsuitable for urgent mode operation. Moreover, the weight of the equipment, carried by zebras was about 1.1 kg, which is also too high in PREDNET project. In [30] an ultra low power WSN for tracking bats in the wild is pro-posed. Small size of the bats strictly limits acceptable maximum weight of the carried nodes to 2 grams including battery. Small battery size limits the amount of available energy in the node. To meet the energy limitations and lower the energy consumption of the nodes, a low power wake-up receiver is integrated in the carried by the bats devices. The collected data are sent to the ground base station when a bat carrying the node flies close enough to this latter. When it happens, the wake-up signal (which is sent periodi-cally by the base station) is received by the low power receiver in order to activate the main transmitter and send the data. Small communication rage with ground stations (about 50 m) leads to spontaneous communications in the system. Relatively high mobility of the bats along with small mon-itoring area (comparatively to the Kruger park) make the communication opportunity happens often. However, as it will be shown in Section 4.3.4, impossibility to cover all the habitat area in the Kruger park and relatively low mobility of rhinoceros will cause high delays for message transmis-sions, which is unacceptable for the alarm messages in the framework of the PREDNET project. Similar application for flying foxes (fruit bats) is described in [96]. The authors propose a 3-tier system containing mobile nodes installed on the animals, gateways and cloud service. In this case, the multihop commu-nications are not provided which, as we will show in Chapter 4, makes impossible rapid delivery of alarm messages from uncovered by gateway zones. However, authors propose to adapt the behavior of the nodes de-pending on the current state of charge which can evolve due to both, en-ergy consumption and energy harvesting. This option can be useful in the framework of PREDNET project. In [103] an inverse-GPS tracking system for birds and small mammals is proposed. The system allows localizing effectively the animals within large areas. However, the project is not focused on data collecting and transmis-sions, which is required by the PREDNET project. Moreover, redundant coverage of the area is required in the proposed system, which, as we will show, is not always possible in the Kruger park.
Another project proposes a solution to monitor migrating whooping cranes[8]. This solution uses less sensors and has less strict constraints in terms of delivery period (it has to be less than 24 hours, which is much bigger than 15 minutes required in PREDNET project). The size limitations are also different due to the difference of sizes of animals. Also, in this project the authors propose to use GSM cellular technology with another short range 802.15.4 radio to send the data. These solutions are not suit-able for PREDNET project due to the short range of 802.15.4 technology and high power consumption of GSM. Moreover, GSM does not allow to manually adjust the communication parameters. The problem of GSM for PREDNET project is related also to not full coverage of Kruger park zone by cellular networks.
Some wildlife animal tracking solutions are based on the satellite tech-nologies (e.g., Argos, Iridium, Globalstar) [40], [28]. Even though these solutions ensure a good coverage of the target area, they have important drawbacks. First, the hardware installed on the animals is very expensive, which makes impossible to equip all the animals in the target area. Second, the energy consumption of these devices is high. That leads to the short lifetime of the nodes, and, thus, makes this kind of solutions unsuitable for the PREDNET project.

READ  Instrumentation for a harsh environment 

PWAN networks

As mentioned in Section 1.3.1, one of the peculiarities of the PREDNET project context is the huge target area. The proposed solution has to en-sure packet delivery from any position of the entire southern part of the Kruger park. Thus, well known in the WSN short range solutions (e.g., IEEE 802.15.4 based transceivers) will not be applicable in the framework of the PREDNET project due to the small coverage. Indeed, with a given node density (1.7 animals per square km), the use of short range commu-nication can cause a high number of isolated nodes (the nodes which are not in range of any other node). In this case, the delivery of the urgent alarm messages form the animal in danger to the BTS will be impossible even with multihop strategy. Thus, long range communication technology ensuring extended coverage is required. As mentioned before, cellular net-work (GSM, NB-IoT, LTE-M) and satellite based solutions are not suitable for the project due to both, high energy consumption and the hardware costs. So, the LPWAN technology [87],[54] seems to be the most appropri-ate to meet functional requirements of the project. Some examples of the LPWAN technologies will be presented in the next section.

Table of contents :

Remerciements
Abstract
Résumé
Contents
List of Figures
List of Tables
List of Abbreviations
1 Global introduction 
1.1 Context
1.1.1 Internet of Things (IoT)
1.1.2 Wireless Sensor Networks (WSN)
1.1.3 Wireless Sensor Node structure
1.1.4 Particularities of RF communications which can impact
the energy consumption
Interference
Collisions
Fading
Other factors
1.2 WSN communication stack
1.2.1 Physical layer
1.2.2 Data Link Layer
1.2.3 Network Layer
1.2.4 Transport Layer
1.2.5 Applicatoin Layer
1.3 Particular WSN application usecase: LIRIMA PREDNET project
1.3.1 Functional requirements
1.3.2 Other wildlife tracking projects
1.4 LPWAN networks
1.4.1 Sigfox
1.4.2 Ingenu (ex. RPMA)
1.4.3 LoRa
1.5 Contributions of this thesis
1.6 Structure of the thesis
2 Experimental study of interference impact on the energy consumption in WSN 
2.1 Chapter introduction
2.2 Motivation and Related works
2.3 Description of the platform
2.4 Experimental setup
2.4.1 Ideal case scenario
2.4.2 Real case scenario
2.5 Measurement results
2.5.1 Measurement approach
2.5.2 Ideal case
2.5.3 Real case
2.5.4 Measurement analysis TX side RX side
2.5.5 Lifetime evaluation depending on RSSI
Ideal case
Real case
2.5.6 UsingWi-Fi as a interfering network
2.6 Long term distributed interference measurements
Description of the platform
Experimental setup
Results
2.6.1 Discussion
2.7 Chapter conclusion
3 Thompson Sampling based Cognitive ratio solution for Multihop WSN 
3.1 Chapter introduction
3.2 Related works
3.3 Optimal Algorithms for Multi-Armed Bandit
3.3.1 Channel selection modeling
3.3.2 Adaptation of UCB approach
3.3.3 Adaptation of -greedy
3.4 Thompson Sampling Based Learning Algorithm
3.5 Performance evaluation
3.5.1 Simulation based on generated channel samples
3.5.2 Simulation over the real channel measurements
3.5.3 Experimentation via real WSN hardware implementation
3.5.4 Evaluation results
3.6 Multihop extension
3.6.1 Test application scenario: EWSN 2016 dependability competition
3.6.2 Proposed multihop solution based on Thompson sampling approach
3.6.3 Performance evaluation results and discussion
3.7 Chapter conclusion
4 Distributed adaptive MAC protocol for WSN-based wildlife protection
4.1 Chapter introduction
4.1.1 Existing MAC protocols
4.2 Our contribution : theWildMAC protocol
4.2.1 General Time division structure
4.2.2 Timeslot structure Control part
First and Second TX/RX parts
4.2.3 Multichannel operation structure
4.2.4 Timeslot reservation and channel access
4.2.5 Communication example
4.3 Performance evaluation
4.3.1 Chosen LoRa communication parameters
4.3.2 Range test and coverage simulation
Range test
Coverage simulation
Obtained results
4.3.3 Cell capacity estimation
4.3.4 Kruger park coverage estimation
4.3.5 WSNet implementation and simulation results
One hop scenario
Multihop scenario
4.4 Chapter conclusion
5 Conclusions and perspectives 
5.1 Conclusions
5.2 Impact of the thesis
5.3 Perspectives
Bibliography

GET THE COMPLETE PROJECT

Related Posts