TECHNICAL REPORT
| Grantee |
Sejong University
|
| Project Title | Assisting Natural Beekeeping in Rural and Remote Areas using LoRa-based IoT and Drones |
| Amount Awarded | USD 30,000 |
| Dates covered by this report: | 2024-01-01 to 2024-06-30 |
| Economies where project was implemented | South Korea |
| Project leader name |
Nhu-Ngoc Dao
|
| Project Team |
Teshager Hailemariam Moges
Fayshal Ahmed
|
Project Summary
This project developed proof of concept ideas for providing Internet availability to efficiently assist beehive management in natural beekeeping in rural and remote areas, where the Internet is unserved. The system includes several IoT devices located at beehives, which connect to a drone through LoRa wireless communications. Periodically, a drone flies on preprogrammed trajectories to collect data from IoT devices. The system is expected to helpfully support beekeepers with sufficient information for hive remote management without human intervention.
The project implemented and configured a programmable drone equipped with a Raspberry Pi board. The drone works properly and can connect with IoT devices through built-in and external WiFi connections. In addition, the project developed an optimal trajectory algorithm for the drone to minimize the mission time by using adaptive clustering and deep reinforcement learning methods. To prove the applicability and performance of the proposed solution, the project set up a simulation environment consisting of a drone and numerous IoT devices, which are randomly distributed on the ground. The drone departs from a predefined station and takes a round trip to collect data from all IoT devices, then comes back to arrive at the station. The simulation results illustrated the outperformance of the solution compared to several existing approaches.
Table of Contents
- Background and Justification
- Project Implementation Narrative
- Project Activities, Deliverables and Indicators
- Project Review and Assessment
- Diversity and Inclusion
- Project Communication
- Project Sustainability
- Project Management
- Project Recommendations and Use of Findings
- Bibliography
Background and Justification
In Vietnam, honey beekeeping has thrived for a long time thanks to its advantages in natural conditions such as tropical climates and forests. According to the Vietnam Beekeepers Association in 2021, there have approximately 1.5 million bee colonies managed by 34,000 beekeeping households, of which 6,350 are professional beekeepers. In particular, natural beekeeping has significantly contributed to the improvement of living standards for local residents in rural areas owing to high economic efficiency [1].
However, beehive management and monitoring in natural beekeeping are challenging because the beehives are dispersedly deployed in rural and remote areas where the terrains are difficult for transportation and telecommunication. Currently, this work requires a huge volume of manpower and time consumption, especially in bad weather conditions. Recent studies showed that advanced technologies can be one of the key solvers for this problem. In particular, IoT devices equipped with various sensors may help to automatically monitor the beehive conditions such as thermal, humidity, bee counter, swarming sound, and beecomb's weight [2]. This information represents the health and quality of the beehives. Via telecommunication systems, professional beekeepers may acquire this information for remote beehive management [3].
To address the difficulty in managing beehives in widely mountainous areas, the proposed system would benefit local beekeeping communities in northern mountainous provinces in Vietnam by facilitating them with remote beehive management abilities. It is expected that jointly utilizing IoT, long-range wireless communication, and drone technologies will help professional beekeepers simultaneously monitor a large number of beehives efficiently and, therefore increase their product quantity and quality.
Project Implementation Narrative
The overall project goal is to facilitate local beekeeper communities with advanced technology utilization for remote beehive management and monitoring. To this end, the objectives of this project are to design a LoRa communication system involving a drone and several IoT devices, which helps to collect necessary information on beehives remotely and automatically. The system design includes an optimal trajectory configuration algorithm for the drone to minimize flying time and improve communication efficiency.
In this project, we have done the following activities:
- Investigate current technological approaches for efficient beehive monitoring based on IoT, drones, and long-range low-power wireless communications.
- The investigation has been done through a literature review of academic and public documents such as scientific research papers and technical reports from reputable databases, e.g., IEEExplore and Elsevier SCOPUS. Through the literature review, we recognized that recent low-power WiFi technologies are considered a potential candidate for the communication between IoT devices and the drone in our system.
- On the other hand, we organized a meeting with local agencies and organizations to gather practical experience, requirements, and expectations. The meeting was held from April 17-22, 2023, in Nha Trang, Vietnam. Attending institutions include Vietnam National University of Agriculture-VNUA, Hanoi University of Mining and Geology-HUMG, Sejong University-SJU, Institute of Veterinary Research and Development of Central Vietnam-IVRD, and local government communities. The meeting minutes are attached in the Documents section. Through the meeting, we recognized that the system should use popular and cost-effective technologies for easy operations and maintenance. WiFi technologies have been recommended if applicable. Another important expectation is low energy consumption and portable energy supply for IoT devices.
- System modeling and solution development
- Based on the investigation results, we decided to model the system including a programmable drone and several IoT devices. In particular, the drone was constituted of a drone frame, pilot processing unit CubePilot Cube Orange+, GPS receiver Here3, radio controller RadioMaster Tx16s Mk2, Raspberry Pi 4 Model B, batteries, WiFi transceivers, and storage cards. In the drone, our proposed trajectory algorithm is installed on the Raspberry Pi board. Regarding the IoT devices, we selected Orange Pi Zero 2W, which supports an integrated WiFi transceiver. The drone and IoT devices have been configured and operated properly.
- We analyzed and modeled system characteristics to optimize the trajectory configuration for the drone. Considered constraints include IoT devices' location, the maximum number of IoT devices in a cluster, the maximum and minimum altitudes of the drone, and the data volume of IoT devices. The system analysis is described in detail in the technical report in the Supporting document section.
- We formulated the objective function to minimize the mission time while ensuring the aforementioned constraints. The objective formulation is described in detail in the technical report in the Supporting document section.
- We developed a two-stage algorithm to resolve the above objective function. First, we used the adaptive K-mean clustering algorithm to assign every IoT device into clusters. Then, in each cluster, we specified the hovering point of the drone by using a deep reinforcement learning algorithm to obtain the minimum transmission time between IoT devices and the drone. The Deep Deterministic Policy Gradient (DDPG) algorithm has been selected to find the optimal continuous values in the objective function. Pseudocode of the algorithm is described in the technical report in the Supporting document section.
- Simulation-based performance evaluation
- We used an NVIDIA GeForce GTX 1650 GPU and an Intel(R) Core(TM) i5-9400 CPU as a computational system in the simulation experiments. The software environment is programmed with Python 3.10.13 (64-bit), PyTorch 1.13.1 with CUDA 11.6 support, and Gym 0.26.2. In Python and Gym, we implemented the system environment of a drone and 100 IoT devices distributed randomly in a 200x200m2 area. The drone is assumed to be moving at a constant speed of 15 m/s. The deep reinforcement learning network, DDPG algorithm, has two hidden layers, the first of which has 512 nodes and the second of which has 256 nodes. The batch and buffer sizes are set to 16 and 50000, respectively. These parameters have been selected based on experimental trials. It is worth noting that other parameters of the DDPG algorithm may be possibly used depending on the hardware capacities.
- To evaluate the proposed algorithm, we measured the system performance in terms of age of information (AoI), which reflects the total duration from the time the data is generated at IoT devices until it is collected by the drone, including the flying time of the drone from the station. With different IoT data volumes and generation timestamps, the proposed algorithm outperforms existing algorithms by 132% compared to a manually random control. Details are described in the attached technical report.
Project Activities, Deliverables and Indicators
Beginning of Project
| Activity | Description | #Months |
|---|---|---|
| Literature review | Investigate current technological approaches for efficient beehive monitoring based on IoT, drones, and long range low power wireless communications. | 2 |
Middle of Project
| Activity | Description | #Months |
|---|---|---|
| System modeling | (1)- Clarify the system model including devices, interfaces, connections, internal and external constraints. (2)- Design the operational algorithms such as IoT data compression, transmission scheduling, and drone trajectory optimization. | 5 |
| Solution development | (1)- Develop the entire system architecture and operations. (2)- Program the proposed algorithms. (3)- Evaluate the proposed system architecture and algorithms in simulation framework. | 6 |
End of Project
| Activity | Description | #Months |
|---|---|---|
| Trial implementation | (1)- Install the algorithms on IoT devices and drone. (2)- Physically deploy the system at northern moutainous provinces in Vietnam. (3)- Monitor the performance and efficiencies of the system. | 3 |
| Result discussion and report writing | (1)- Write the final report and SCI(E) paper. (2)- Specify next stage after the project such as large-scale deployement and research extension. | 2 |
Key Deliverables
| Deliverable | Status |
|---|---|
| Trajectory algorithm to be completed by July 2024 | Completed |
| System design to be completed by July 2024 | Completed |
Key Deliverables - Detail
| Deliverable: Trajectory algorithm to be completed by July 2024 Status: Completed Start Date: September 1, 2023 Completion Date: June 30, 2024 Baseline:Currently, the trajectory of drone is controled by either a pre-configured algorithm or a manual control. These methods cannot optimally adapt to the current traffic demands of IoT devices. Hence, the flying duration takes long time for a roundtrip to collect all information from the IoT devices. Activities: - We analyzed and modeled the system and environment characteristics. - We developed an algorithm using adaptive K-mean clustering and deep reinforcement learing method to minimize roundtrip flying duration of drone. Outcomes: Algorithm to decide optimal drone's trajectory Additional Comments: An optimal trajectory algorithm is designed to decrease the roundtrip flying duration of drone compared to those of pre-configured and manual control algorithms. |
| Deliverable: System design to be completed by July 2024 Status: Completed Start Date: January 1, 2023 Completion Date: October 31, 2023 Baseline:- Using human resources to manually visit and perceive the conditions of bee hives. - Using short-range wireless or wireline connection to collect sensing information from IoT devices mounted at bee hives. Activities: - Investigate appropriate LoRa transmission technology for the system through literature review and a meeting with stakeholders. - Integrate necessary modules to build a programmable drone and several IoT devices. - Configure the drone and IoT devices for communications. Outcomes: A system consists of a programmable drone and several IoT devices, which are connected using low-power WiFi technologies. The cost of one drone is $1,056.93 and that of one IoT device is $31.26. For comparison, a commercial drone with similar functions is Parrot Anafi AI costed $4,500 [4]. Note that currently, we have not yet identified any application of drone in farming monitoring in Vietnam. Additional Comments: Develop a system that is constituted by a drone and multiple IoT devices. The IoT devices connect to the drone using wireless communication such as built-in wifi or attached long-range wireless interfaces. |
Project Review and Assessment
We are performing the project as planned without significant changes in activities as well as schedule. We have proved the efficiency and feasibility of the concept of utilizing drone and IoT devices to support native beekeeping.
The findings open future opportunities to extend this study to implement the system in real scenarios, not only to support nature beekeeping but also in various agriculture and forest contexts. Based on the findings of this project, we intend to apply for a scale up grant next year.
While working with local beekeepers in Vietnam and the government officers, we recognized that several aspects should be considered for upcoming studies such as power supply for IoT devices, physical protection of the devices, and the impact of weather on the system operation. These aspects will be further studied in the next phase of this study.
Diversity and Inclusion
This project has been studied by a professor and two MS students, who are with different levels of knowledge and cultures. In addition, we closely collaborated with local Vietnamese agencies to investigate the requirements and opinions of local stakeholders to efficiently design the system and solution.
Regarding the communities served, the project facilitates beekeepers with convenient tools for remote beehive management. Hence, women and people with disabilities can easily do the work that men do.
Project Communication
We held a meeting with local Vietnamese partners to collect practical experiences, requirements, and expectations as well as an understanding of deployment challenges.
We participated in the KICS Winter workshop to meet researchers in the field and share our findings and experiences.
Project Sustainability
We held a meeting with local Vietnamese partners to collect practical experiences, requirements, and expectations as well as an understanding of deployment challenges.
Through the project, we identify open opportunities to extend our proposed system to support other communities and applications such as search and rescue, forest fire detection, weather forecasting, remote sensing, hiking, and trail running.
Project Management
Sejong University has dedicated departments to support and manage the agreement, purchase, and accounting tasks. We follow the policies of APNIC and ours for these activities.
From a technical management aspect, the team includes one Assistant Professor and two MS students.
In addition, we received support from local Vietnamese partners during the project. This project helps to develop our international collaboration.
Project Recommendations and Use of Findings
We closely collaborate with local Vietnamese partners in any activities of the project to ensure that our development aligns to the requirements and expectations of potential stakeholders.
Bibliography
[1] Vietnam Beekeepers Association, http://www.vba.org.vn
[2] Tashakkori, R., Hamza, A.S. and Crawford, M.B., 2021. Beemon: An IoT-based beehive monitoring system. Computers and Electronics in Agriculture, 190, p.106427.
[3] Arshad, J., Aziz, M., Al-Huqail, A.A., Zaman, M.H.U., Husnain, M., Rehman, A.U. and Shafiq, M., 2022. Implementation of a LoRaWAN Based Smart Agriculture Decision Support System for Optimum CropYield. Sustainability, 14(2), p.827.
[4] Parrot Anafi AI product, https://www.floridadronesupply.com/products/parrot-anafi-ai
See also the attached technical report.
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License 
