INTelligent Edge using Learning Loops & Information GEneration for NeTwork State Inference-based Automation – INTELLIGENTSIA
Differentiated Quality of Service in LoRaWAN: An Efficient and Low-Cost Solution for High-Performance IoT
LoRaWAN is known as an efficient and cost-effective solution for the Internet of Things (IoT). However, this technology has limitations in terms of performance. As IoT becomes a major economic priority for various stakeholders (local collectivities, service providers, etc.), the project aims at developing a set of methods to improve LoRaWAN's performance, particularly regarding quality, such as ensuring reliable data packet transmission.
To cope with strong growth in the number of connected devices and the quality requirements of new services, the project aims at developing various tools for differentiated quality.
The project aimed to enhance the attractiveness of the LoRaWAN network, a protocol widely used in IoT communications, by developing differentiated quality of service tailored to the specific needs of professional clients, particularly those of Orange. This initiative sought to overcome the limitations inherent to the CSMA/CD protocol, often seen as a bottleneck in contexts requiring high performance and high terminal density. One of the project's objectives was to introduce differentiated quality into LoRaWAN. The focus was on creating a LoRaWAN infrastructure capable of delivering quality levels adapted to demanding use cases, such as container management for logistics or other critical services for professional clients. A second objective involved network automation and reconfiguration. The development of automation and reconfiguration mechanisms enabled better network resource management, notably by optimizing the dynamic allocation of channels to address demand spikes or specific constraints. The final objective was to enable massive IoT terminal support. Efforts were made to extend the network's capacity to handle a large number of simultaneous terminals, thus meeting the growing demands of complex IoT ecosystems. The project’s ambitions went beyond technical improvements. A key challenge was to make LoRaWAN appealing for services requiring high-quality service to attract professional sectors with specific needs, such as container management or asset tracking in environments where reliability is critical. Another challenge involved exploring new application domains, such as developing innovative services in sectors like road and maritime safety, where reliable and rapid communication can directly impact human lives and infrastructure management. A more technical challenge was to study the technological limits of LoRaWAN, particularly those related to its intrinsic constraints due to the CSMA/CD protocol, which can cause collisions and reduce efficiency in heavily utilized environments. These efforts have led to potential improvements, paving the way for a new generation of IoT services based on LoRaWAN.
The implementation of the project was based on a combination of mathematical and simulation tools, technological developments, and practical testing. This approach aimed to address the project's technical and strategic challenges while ensuring the achievement of defined objectives.
The foundation of the project was based on precise mathematical modeling of communication mechanisms and terminal interactions within a LoRaWAN environment. This allowed for the design of optimized algorithms to enhance network performance, particularly in terms of quality of service (packet delivery ratio) and resource management. To validate the theoretical calculations, simulation and emulation tools (notably ELoRa) were developed to analyze network behavior in various scenarios. These tests enabled performance predictions, bottleneck identification, and algorithm validation.
The hardware and software design of LoRaWAN systems within the project was carried out by Aguila. This phase included the development of LoRaWAN-compatible equipment and the creation of a monitoring interface providing real-time feedback on network and terminal performance. Specific test suites were developed to validate the performance of the designed systems, particularly in the context of a bike fleet management project. This initiative aimed to provide partners with concrete data and relevant use cases to assess the effectiveness of the proposed solutions.
The project achieved highly interesting results, both theoretical and practical, highlighting the potential for improving LoRaWAN to address complex and diverse requirements. These advances covered theoretical aspects of mathematical modeling, simulation tools, and practical solutions implemented in real-world scenarios.
1. Algorithms for Differentiated Quality of Service
One of the project's major accomplishments was the development of optimized algorithms for medium access. These algorithms enable:
- Prioritizing specific data flows based on the needs of critical applications.
- Reducing collisions in highly utilized network environments.
- Enhancing network reliability and performance for demanding use cases, such as container management or industrial asset tracking.
These innovations position LoRaWAN as an effective and cost-efficient solution for professional clients with high quality of service requirements.
2. Development of a LoRaWAN Network Emulator for Large-Scale Deployments
The project led to the creation of an emulator capable of modeling large-scale LoRaWAN networks. Unlike demonstrators limited to a small number of devices, this tool allows:
- Simulation of massive networks comprising thousands of devices to analyze the performance and limitations of differentiated quality protocols.
- Evaluation of the impact of new algorithmic solutions in realistic scenarios before field deployment.
- Provision of a flexible environment to test various network configurations and fine-tune operational parameters.
3. Deployment and Data Collection in Real-World Systems
The project also included the deployment of various LoRaWAN systems on bike fleets, illustrating a practical application of the developed solutions. These systems enabled:
- Data collection under real-world conditions, enriching the understanding of network performance in dynamic environments.
- Providing valuable insights to project partners for adjusting their own solutions and assessing practical use cases.
The results obtained within the framework of the project open the way to numerous opportunities for improvement and application, both technologically and commercially. Two major axes are emerging for the next steps. The first concerns a potential collaboration with ChirpStack to study the integration of algorithms developed within the project, ChirpStack being a key player in open-source solutions for LoRaWAN networks. This initiative aims to present and discuss the algorithms developed during the project, in particular those aimed at improving access to the medium and offering a differentiated quality of service. One target is to integrate these algorithms into ChirpStack code, in order to benefit from broader adoption by the LoRaWAN community and allow other users to take advantage of the advances made. This also makes it possible to strengthen the synergy between the problematic technical solutions of the project and the standard network management tools, favoring their deployment on a large scale.
A second axis is the implementation of the solution for the maritime security market in West Africa. A promising market has been identified where LoRaWAN technology could play a crucial role in maritime security systems. These perspectives include both the detection of marine pollution, allowing precise and real-time monitoring of environmental incidents, while quickly alerting the competent authorities, and the strengthening of the security of maritime zones, thanks to the use of LoRaWAN networks to monitor activities and detect potential risks. A strong point of LoRaWAN is its low energy consumption and its ability to cover large areas, key assets in environments where communication infrastructure is sometimes limited.
In the era of network softwarization, INTELLIGENTSIA objective is to fast move towards automation of end-to-end network operations by leveraging advanced learning algorithms able to scale with large-scale virtualized networks and to meet control and operations requirements of massive IoT use-cases. The envisioned scientific contributions are related to the elaboration of a dynamic state machine framework for modeling a network virtualization platform, based on clustering and classification algorithms against data sources related to traffic, device, servers and virtualized network functions and services. The learning framework is meant to discover in real-time known and unknown network states, as well as to anticipate reconfigurations of the network and in particular of novel IoT radio access functions besides device behavior and network backhauling. The expected impact includes the enhancement of network automation platforms under consideration by the industry and the design of novel IoT radio access functions.
Project coordination
Fabrice GUILLEMIN (ORANGE (Orange Labs -Gardens))
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
ORA ORANGE (Orange Labs -Gardens)
AGUILA SAS AGUILA TECHNOLOGIE / Operations
Inria Rennes - Bretagne Atlantique Centre de Recherche Inria Rennes - Bretagne Atlantique
CEDRIC CENTRE D'ETUDES ET DE RECHERCHE EN INFORMATIQUE ET COMMUNICATIONS
Acklio / Recherche
Help of the ANR 772,370 euros
Beginning and duration of the scientific project:
November 2020
- 42 Months
