Deposit code: PN-III-P3-3.5-EUK-2016-0009
Cod proiect 9 950
Contract number 86E din 01/09/2016
The total value of the budget: 918.000,00 Lei
The total value of the contract: 1.387.400,00 Lei
The total value of the co-founding: 469.400,00 Lei
The start date of the contract: 01/09/2016
The end date of the contract: 31/08/2019 [31/08/2022 – project ending date] Contract duration: 36 luni
Project duration: 72 luni
Coordinator: BEIA Consult International SRL
Partner: Seacon Europe Ltd
Project Director (coordinator): Dr. Ing. George Suciu
The intelligent forest protection monitoring system envisions to materialize and implement an integrated wireless sensor network (WSN) concept for detection, tracking, and confirmation of random events with potentially destructive effects on the forest environment. The monitoring solution will be designed to reduce environmental vulnerabilities of the forest by anticipating trends and intensities of threats (such as vehicles, pests, fire, extreme weather), and working with emergency services to reduce the effects of disasters.
The main reason for the project is the safeguarding of natural environment and prevent landslides, especially in forests around the Danube and tributary rivers. The system is built on a robust architecture that integrates a WSN network of ground sensors, central server, the radio and wired communications, and a layer of intelligence information system. Reliability and security of the system are based on special measures of physical protection of the system components exposed to corrosion of the environment, energy efficiency in the arrangement of sensors, network topology adaptation of sensors and radio propagation constraints in the forest environment, and multi-modal methods of secure access to databases.
Furthermore, we will implement a unique solution for monitoring objects (such as vehicles, electrical machinery or motors) by identifying them primarily with analysis of sound and/or vibration measurement data. Expected outcomes of the project are maps with the location and frequency of events, databases, geospatial response plans to events, applications for storing, processing and fusion of data in a functional model of the system. The project introduces three innovative components in the system, namely energy efficiency in forest environments for increasing the life of the network of sensors, then a statistical model of risk factors and threats from the forest, used to predict and confirm an event, and collaborative automation of system resources and intervention services in case of environmental threat events.
The system will be sold in a basic scenario as a nationwide SaaS service to stakeholders in the forestry domain, especially for efficient monitoring of objects, disaster early warning management and related disaster forecasts. In an extended scenario, the commercialization of the services of the platform will be enlarged in the Balkan/Danube region through BEIA’s sales and partners network.
We describe the motivation behind the SeaForest project, the problems related to the system architecture, its reliability and security, and also the innovative components mentioned above. We refer to all of the characteristics indicated, by performing the related activities in the following phases.
Phase I – State of the art analysis
Activity I.1 Technical and scientific state of the art
This Phase I of the SeaForest project concluded in December 2016. It contained only one activity which aimed to define the actual knowledge state in the sensors field and the possibility to organize them in networks. The research purpose of this phase consists of studying both the sensors and the intelligent forest monitoring systems, which will reduce the impact reduce the impact that various factors may have on the forest environment.
Phase II – Defining the monitoring system
Activity II.1 Market analysis and user requirements
Activity II.2 Use case definition
Activity II.3 Designing the system architecture
In Phase II we analysed the target market for the proposed solution in terms of size, trends, user requirements, similar solutions with applicability in the target area, and also have been studied the barriers to implementing an effective forest management strategy. The proposed solution has been presented in relation to the market to which it is addressed, the competitive advantages that it could offer, the stakeholders who might be interested in the solution or which could facilitate its capitalization on the national and international market. Also, within this activity, we presented the use cases and designed the system architecture. Phase II concluded in May 2017. After defining the monitoring system, we are in the course of analyzing the techniques and methods for the system development.
Phase III – Techniques and methods for the system development
Activity III.1 Sensor context analysis
Activity III.2 Developing the communication solution
Activity III.3 Methods for data processing
Phase 3 was completed in November 2017.
In this phase, for a proper later design of the SeaForest system, the team has investigated the main standards and regulations specific to the silvic environment, such as the international PEFC regulations (Programme for the Endorsement of Forest Certification) and Regulation (EU) No 995/2010 of the European Parliament and of the Council of 20 October 2010 laying down the obligations of operators who place timber and timber products on the market Text with EEA relevance.
We have devised the functional and development specifications for the software components and the telecommunication solution. After research, the team decided to choose the single-board computer Raspberry Pi 3 B and a set of compatible sensors as the acquisition and processing environment. They have been selected for their flexibility and capacity to accommodate to a broad range of practical issues in domains connected with environment protection and preservation.
Phase IV – Design and test the remote sensor solution
Activity IV.1 Design the remote sensor solution
Activity IV.2 Realization and integration of the remote sensor solution
Activity IV.3 Testing the remote sensor solution
Phase 4 concluded in May 2018. The main goal was establish a workable configuration for the SeaForest monitoring system by testing the behaviour of several sensor solutions (previously introduced), in what concerns data acquisition and connectivity issues. One of the solutions considered includes Raspberry Pi 3 Model B. A suitable configuration was proposed, and the requisite devices to be connected to the board such as microphone, analog – digital converters, etc. were investigated. A framework was developed for signal acquisition and pre-processing.
A second objective of this phase was to choose an appropriate methodology to differentiate several types of environmental noises. We tested a range of acoustic signal pre-processing, processing and modelling techniques. The sound database consisted of four distinct types of noise: chainsaw, vehicle, genuine forest, and a forth class of complementary environmental noise. It served to evaluate several features such as zero crossing rate, signal energy, Mel-frequency cepstral coefficients, linear prediction coefficients. In what concerns feature space modelling, we have assessed two modelling approaches: SVM (Support Vector Machines) for binary classification and Gaussian, Mixture Modelling (GMM) for multiple class classification, using either closed-set or open-set sound identification. The SeaForest team has presented the results of these experiments in two important dissemination events in Romania.
Phase V – Development of the software platform; Integrate, test and validate the SeaForest system
Activity V.1 Interfaces and connectors implementation
Activity V.2 Development of the data monitoring and analysis module
Activity V.3 Development of the notification module
Activity V.4 Defining the testing and validation scenarios
Activity V.5 Testing and Assessment of the integrated system
Activity V.6 Dissemination and Market exploitation
Phase V was completed in December 2018.
After concluding Phase V, the following 3 phases will report the economic and financial activity as follows:
Phase VI – report the economic and financial activity till December 2019.
Activity VI.1 Defining the system testing and validation scenarios
In this activity, starting from the description of the general operating mode of the system and its interaction with the user, the test plan and scenarios are presented.
Activity VI.2 Testing and evaluation of the integrated system
The objective of this activity is to evaluate the system’s performance, which could be a justification for the chosen solution.
Activity VI.3 Market dissemination and exploitation
Phase VII – report the economic and financial activity till December 2020.
Phase VII – report the economic and financial activity till 2021.
Papers resulted from the project:
- Suciu G., Ciuciuc R., Pasat A., Scheianu A. (2017) Remote Sensing for Forest Environment Preservation. In: Rocha Á., Correia A., Adeli H., Reis L., Costanzo S. (eds) Recent Advances in Information Systems and Technologies. WorldCIST 2017. Advances in Intelligent Systems and Computing, vol 570. Springer, Cham
- G. Suciu, E. Olteanu, G. Todoran, V. Suciu, A. Scheianu, ” Wireless Sensor Network and Cloud Platform for Education in Forest Monitoring and Protection”, SIITME2017
- E. Olteanu, V. Suciu, S. Segarceanu, I. Petre and A. Scheianu, “Forest Monitoring System Through Sound Recognition,” 2018 International Conference on Communications (COMM), Bucharest, 2018, pp. 75-80.
- Elena OLTEANU, Svetlana SEGARCEANU, Inge GAVAT, ENVIRONMENTAL NOISE CLASSIFICATION USING THE GMM APPROACH SISOM & ACOUSTICS 2018 , May 24-25 2018, Bucharest, to appear in online and in Acta Electrotehnica.
- George Suciu, Ana Petrache, Cristina Badea, Ijaz Hussain, Tony Buteau, David Schachet, Loic Durand, Matthieu Landez, “Low-Power IoT Devices for Measuring Environmental Values”, SIITME2018.
- Alina Elena Marcu, George Suciu, Elena Olteanu, Delia Miu, Alexandru Drosu, Ioana Marcu, “ IoT System for Forest Monitoring ”, TSP 2019, Budapesta, Ungaria
- Elena Olteanu, Svetlana Segarceanu, Delia Oana Miu, George Suciu, Alexandru Drosu, Inge Gavat, “Fusion of speech techniques for automatic environmental sound recognition”, Timișoara, România (in process)
- Svetlana Segarceanu, Elena Olteanu, Inge Gavat, “ Evaluation Of Speech Specific And Non-Speech Classification Techniques For Environmental Sound Recognition ”,București, România.