1. Home
  2. Teams
  3. Autonomous Networks & Environments (ERA)

Autonomous Networks & Environments

The ERA team is a spin-off of the LM2S team and its research objectives since 2008 have been focused on the theme of autonomous networks. The team includes 1 PU (D. Gaiti), 4 MdC (G. Doyen, M. Esseghir, and M. Lemercier) including one HDR (L. Merghem) since 2010 and 1 ECC (L. Khoukhi) HDR since 2015. The teacher-researchers of the team are all in 27th section and all have or have had significant collective or teaching responsibilities within the institution: current responsibility for an engineering program (ISI), responsibility for teaching programs, co-responsibility for PST (transversal theme STMR), participation in the various advisory and decision-making bodies of the institution (member of the Board of Directors, member of the Board of Directors, member of the Board of Directors, member of the UMR Council and the Board of the doctoral school). It should be noted that all the teacher-researchers (MdC and PU) in the team have the PEDR.
 

Research areas of the ERA team

The team is interested in the management, in the broadest sense, of networks (by collective intelligence techniques for example) with the aim of building an autonomous network, capable of self-configuring, self-repair, self-protection, etc. This research is particularly innovative since it combines techniques from several academic disciplines (networks, artificial intelligence, computing, distributed systems), ambitious since it aims to propose solutions to the complexity of current networks, and vast since it brings together the fields of network architectures, protocols, management and control and the consideration of security issues. This set constitutes what should be the backbone of future networks with highly autonomous characteristics.

The team also focuses its efforts on the notion of services in the broadest sense and by integrating safety into it. From this point of view, the network is no longer a physical architecture whose characteristics are improved but becomes a real support for the services transported and must provide them with quality and protection.

The addition of the notion of services only follows from the notion of an autonomous network. We put our skills not only at the service of the "autonomous network" theme by proposing definitions and implementation of key concepts in the field (knowledge plan, located view,...) but also at the service of certain networks where it seems difficult to do without "autonomy" such as vehicle networks, sensor networks or smartgrids.

Our three directions of research


Auto- piloted networks

At the moment, in order to satisfy the demands of transported applications, several service quality architectures have been proposed, but what they have in common is that they propose a static approach to an environment which is highly dynamic.
The algorithms are often complex and only give good results on very precise data and at particular moments.
Along this axis we have chosen to carry out auto- piloting of algorithms for monitoring and managing various existing networks. The idea is to make the network autonomic.
 

Autonomic networking tools

Under this heading, the approach chosen to pilot the networks comes from techniques of distributed artificial intelligence and mainly from agent systems. The dynamics of the system represented by the agents; decision-making is loaded in the network element which can thus become autonomic and react to a problem. This approach gives a permanent guarantee of satisfaction for the network user and operator. The technology produced can be described briefly as a set of agents which will function in coordination to solve a problem or find a solution in a given situation.
 

Test and simulation environment

 One of the main themes is experimentation of proposals made. Two test and assessment platforms are used. The first is a local environment with standard equipment used for IP, MPLS, or service quality. As for large-scale tests, they are carried out within the Planet-lab network. In this way the test environment is as wide as the Internet. Peer-to-peer network applications are assessed in this framework. Lastly, the team also develops its own simulation and experimentation tools.
The ERA team has developed specialized research in the field of networks and their management and also gained recognition for competence in multi-agent technologies.
This dual skill is indisputably the team’s strength even more so as it is assisted by a complete set of testing and simulation tools.