Cognitive networks and future internet

Convergence of telecommunication networks, mobile networks, TV and Internet leads to evolution of all-IP networks. Classic communications and Internet traffic modelling and traffic engineering techniques have to follow the trend of such evolution.

Traffic modelling and engineering have been playing important roles in network operations and long term plan for global communication infrastructure. This implies the trade of capital expenditure and operational expenditure, and efficient utilisation of network resources, as Internet traffic will fluctuate during hours, days, months and years.

Furthermore, communications systems and networks consume about 3% of electric energy globally. Therefore, in addition to building efficient network infrastructure and cost efficient network operations, accurate modelling of Internet traffic can also help to save electricity by reducing the operational levels of the network resource without degrade the quality of service (QoS) to end users and network applications. Therefore, new traffic models and traffic engineering become important research topic in the future generation of networks.

Network virtualisation has been regarded as a promising solution to enabling flexible utilisation of network resources in open shared network environments. The rationale behind is the newly evolved business relationship between stakeholders, with the separation of various service providers from the underlying network infrastructure provider. In this context, we have been researching efficient resource management schemes based on virtualised network infrastructures. Specific topics include: generic virtual resource management frameworks, virtual resource allocation and adaptation schemes in software-defined networks and virtual network embedding algorithms.

In recent years, research activities on energy efficiency have been targeting future green networks that will substantially reduce energy consumption and CO2 emissions. In this context, the traditional practice of resource over-provisioning and the “always-on” strategy in operational ISP networks is not anymore be regarded as “cool” solutions. We have been addressing energy efficiency issues in heterogeneous network environments, including backbone telecom networks, data centres and content delivery networks, as well as wireless sensor networks.

In addition, we have been researching cross-layer traffic optimization schemes based on interaction between content applications (in particular Peer-to-peer overlays) and the underlying IP-based networks.

As the Internet’s global importance grows and users demand access while mobile, wireless networks have been a major area of research for communications.  Many different types of wireless network have been identified, each with their own technical requirements.  Some of the more important categories are:

• MANETs (mobile ad hoc networks): Generic, self-organising networks where the devices form their own networking infrastructure
• VANETs (vehicular ad hoc networks): Characterised by linear motion (along roads), with access to roadside devices
• WSNs (wireless sensor networks): Typically stationary, often low power and low capability devices, often operating in power-saving mode and hence with only limited connectivity.

Intermittently connected networks (ICNs), sometimes also known as delay- or disruption-tolerant networks (DTNs): networks where an entire end-to-end path between two devices rarely or never exists, due to either a low density of nodes and/or large distances between devices.

Key research topics for these networks include the following, all of which are the focus of research at ICS.  

At the network layer, topology and geographical routing protocols and algorithms have to be developed which take account of each network characteristics such as mobility, node availability and connectivity. QoS requirements vary between the network types and have to be included in the development of routing protocols.
At the transport layer, conventional protocols such as TCP are inadequate in intermittently connected networks. The bundle protocol has been developed for interplanetary DTNs but is not sufficient for terrestrial networks, where transport layer issues such as congestion control have to be considered to avoid overloading nodes.

Security of these networks is important, particularly where the network devices are not under the management of a single entity, as is often the case in wireless networks.  In particular, this makes the networks susceptible to attacks by intruders; yet traditional intrusion detection approaches developed for fixed networks are not suitable for distributed wireless networks.

Finally, green wireless communications are becoming an important topic. This includes energy efficient wireless networks and terminals; and in addition the use of wireless communication technologies for green research such as smart grid, IST, e-health, etc. should also be included.

Cloud computing technologies have been developed rapidly and generated significant impacts on how citizens and businesses use technology and individuals live their everyday lives. In addition to the traditional desktop and laptop computers, smartphones and mobile devices and Internet of Things (IoT) are all connected to the Internet and become part of the cloud.

Due to the new concepts, new services and applications have been developed. The connection of everyday devices, such as home appliances or specialised equipment such medical devices to the cloud technologies lead to our focus of research area.