» version anglaise » RESEARCH » Transversal topics » Future networks

Future networks

In the networking field, the main purposes of current researches and developments are to increase capacity in order notably to handle new traffics and new services, to improve quality of service in terms of both performance and reliability and to support the ubiquity of a global and generalised mobility (network mobility and multi-homing).

The number of available access technologies is growing fast (2G, 3G, 3G LTE, 4G, Wi-Fi hotspots, Wi-Max, xDSL, FTTx, cable, or satellite).

A key concept is the generalisation of self-organisation and autonomic networking, as a mean to create new service paradigms with a user-centric approach or to reduce the operational cost of existing services. Peer-to-peer applications and ad-hoc networks are typical examples of self-organisations at the application or network level. Various Digiteo groups are currently working on several of these topics within several collaborations with telecommunication companies and operators.

Wireless and cellular networksWithin Digiteo research activities in mobile networks area, one important contribution has been OLSR Protocol, which is now widely used within the community of ad hoc networks.

The current research activity on this protocol is the introduction of advanced new features (QoS, multicast and security) and the interworking on the future Internet (super nova) including thing2thing communication, connectivity, auto-configuration and delay tolerant networks.

Wireless networks with or without infrastructure (ad hoc) are under active studies within Digiteo. Technologies based on 802.11x, 802.16x, HSPA and LTE are very important for current and next generation telecommunication networks. Network planning, traffic engineering, optimized MAC scheduling and resource allocation algorithms for OFDMA and SOFDMA are investigated. More globally, cellular planning and traffic engineering based on these technologies remain open domains of investigations. Intelligent Transportation Systems, which will constitute a significant part of the ubiquitous environment, will use these communication technologies to improve the traffic integrity.

Sensor networks are another active field of research where the studies cover the QoS, performance and reliability of these networks. Autonomous addressing and clustering algorithms are two key activities in the domain of sensor networks.
In wireless networks, frequency is the rare network resource. Radio frequency spectrum optimisation is a key objective. In this matter, some teams within Digiteo are investigating the potentialities of so-called agnostic bandwidth utilization. Moreover, one important approach covered by Digiteo projects is the idea of global optimisation within each layer and between layers where a certain independence between these is assumed allowing flexibility in the network utilisation.

The Dynamic Spectrum Allocation (DSA) and spectrum management within the radio system or between heterogeneous radio systems appear to be interesting approaches in order to obtain better spectrum utilisation and more efficient schemes of mobility management. DSA will become available with the development of software radio technologies under studies within Digiteo research groups and are likely to perform very well in dense networks. The potential uncoordinated deployment of cells (eg. Femto-cell) will bring more randomness in the coverage. In such context, opportunistic mobility anchoring solutions, traffic aggregation and prediction based handovers are essential for providing to users strict QoS guarantees especially under stringent traffic conditions.

Automatic networking

Beyond the various topics in wireless networks area presented above, the concept of selforganisation and autonomic networking appears to be an essential and federative theme. Peerto- peer applications and ad-hoc networks are typical examples of self-organisation at the application level (resource management, key distribution), the network level (routing, network coding, security) or the physical level (coding, cooperation).

Optical networks

The current work on ultra high-speed networks is conducted in the field of all-optical networks. Two distinct approaches corresponding to two distinct bandwidth granularities are particularly investigated : optical circuit switching (OCS) and optical packet or burst switching (OPS/OBS). Concerning OCS, two aspects are considered : impairment-aware routing and wavelength assignment and network monitoring. OCS is progressively deployed in core networks.

Concerning OPS, innovative MAC (medium access control) protocols, control plane protocols and routing schemes are developed. OPS /OBS deals with the high volume and short lifetime data transfers. It is considered as a promising technology for future metro/access networks. These studies concern the metropolitan access network as well as the core network.

New-generation access networks are also investigated with a particular focus on hybrid optical-wireless infrastructure. The main objective is to facilitate dynamic spectrum allocation in LTE cellular networks thanks to new backhauling architecture exploiting the potentialities of Radio-over-Fiber technologies.

Cloud networking

An important activity is dedicated to multi-domain Cloud computing.
Scalability, security, trust domains, accounting and billing are some of the major investigated issues within Digiteo teams. For instance, in the context of Cloud over IP networks, innovative traffic metrologies have to be developed in order both to facilitate efficient traffic engineering and fairer pricing metrics.

Advanced modelling techniques

As seen above, network analysis considers quite a variety of aspects in the organisation, evolution, and use of networks. Not surprisingly, numerous approaches are used, and their join exploitation raises interesting challenges. Among them, one can cite graph theory and advanced operational research, various formal description techniques, the whole spectrum of optimisation methods.
Games theory has been used to model the cooperation between selfish operators and is now widely considered for stochastic protocols. Network information theory is used to investigate the performance limits of wireless networks in terms of end-to-end traffic transport capacity theoretical limits. Stochastic modelling and simulation are key tools to analyse and optimise the performance of the network.

Current Digiteo initiatives

Traditionally, most research projects study in depth, and get great results on one of the following so-called network layers : Applications, where specific data compression is the key issue ; Network layers, where the point is allowing the signal to reach the correct access point ; a Medium layer where packets and channels are managed ; and the Physical layer where the actual transmission takes place. The new emerging networks have to deal with various type of signals and various types of constraints (QoS, delays...). They will have to rely on increased cooperation both among nodes and among protocol layers to ensure some global satisfaction of such various requirements.

This work requires a wide variety of knowledge, from theoretical issues such as graph theory or information theory, to system level issues (resource management, incitement to cooperate...), physical layer issues (coding, cooperation) and network layer issues (routing, network coding...). It turns out that Digiteo teams cover this wide spectrum of expertise. For one year, Digiteo has launched and organised the NC4wireless research program with the ambition to progress quickly towards a comprehensive solution to the underlying problems.

A complete description of the program (40 pages) and of its 5 workpackages is available on demand.