Internet and Internetworking

Other Projects

Network Congestion Control and QoS Provisioning

contact: Dr Qiang Fu .

As the network bandwidth increases, it seems congestion control may not be an issue in the future. This may be partially true. But, think about our highway systems. There are always occasions you can drive as fast as the speed limits allow; there are definitely occasions snails would smile at you. Apparently, we need a congestion control mechanism, which is highly adaptive.

Another issue is that the current congestion control mechanisms are designed for the wired networks. They are struggling with the features brought by mobile and wireless networks. These features include non-congestion triggered packet losses, delay variation and mobility.

A third issue is the greedy behaviours of network users. Have you ever set up multiple TCP flows for a given application, or have you ever modified TCP parameters making your TCP flows send data faster. This is essentially "stealing" bandwidth from the competing network users. Unfortunately, the Internet is designed based on the end-to-end principle. It is very hard (if not impossible) to solve these problems under the end-to-end principle.

This project looks at addressing these issues from the perspectives of resilience, scalability, responsiveness, bandwidth utilisation and fairness in various environments.

Concurrent Data Transfer for P2P, Grid Computing, Content Distribution Networks and Others

contact: Dr Qiang Fu .

This project investigates the use of parallel data transmissions to improve network performance. The use of parallel data transmissions could be over the same path or multiple paths and between a single pair of sender / receiver or a set of senders / receivers. One scenario could be that you are sending data through both the Ethernet adapter and the wireless adapter at the same time. Another popular example is BitTorrent, a P2P application. In BitTorrent, multiple concurrent TCP flows are established between a set of senders/receivers over multiple paths. In Grid computing, concurrent TCP flows are used for intensive data transfer. This concept can also be used in Content Distribution Networks (CDNs) and bandwidth-limited wireless networks. There are also many other possible applications. For example, it can facilitate seamless handover in mobile networks, enhance information security, and improve transmission reliability.

Cross Layer Design Using Software Defined Radio

contact: Dr Qiang Fu .

The layered network architecture has greatly helped the evolution of the Internet. But overtime, it has created isolation between solving lower-layer and higher-layer problems. The problems have become more complex and harder to solve in this kind of isolation. There is no doubt cross-layer coordination is the way to go.

This project investigates cross-layer coordination using software-defined radio (SDR). The traditional use of SDR is limited to the domain of physical layer. However, the current development in SDR has made it possible to adapt all the layers in the protocol stack. This presents unique challenges and opportunities for the design of cross-layer mechanisms.

IP Networks for Efficient Energy Generation, Consumption and Distribution

contact: Dr Qiang Fu .

"green" has been a hot word for politicians. It is equally hot for academics. The Information and Communication Technology (ICT) consumes approximately 2% of the global energy consumption. Making an all-optical network would save energy because the routers do not need to maintain a packet queue. Making ICT devices more intelligent would save energy when the users are not active, the device could switch to energy-saving mode. However, all the reduction is within the 2%. It is not much. How about use ICT to help other sectors save energy.

This project investigates the possible network architecture and protocol suite that enables intelligent scheduling of energy generation, consumption and distribution. For example, the network collects real-time energy demand and sends the information to the generator network, which adjusts its energy generation scheduling accordingly. In the opposite direction, the generation network may advise a community network, suggesting stop non-urgent energy consumption if possible. The community network may adjust its consumption behaviours accordingly. Another issue is how to distribute the energy from the generator network to the community network. There would be multiple distribution paths %u2013 it sounds like a routing problem.

Network Security in Wireless Sensor Networks

contact: Dr Qiang Fu .

This project investigates the attacks in Wireless Sensor Networks (WSN) and the security issues for data aggregation, routing, localization and key management.

Localisation in Wireless Sensor Networks

contact: Dr Qiang Fu .

Have you used or heard about location-aware applications. When you are travelling around, your hand-held smart phone tells you what places of interest are nearby. It may even be able to make a schedule for you. This would help you fully enjoy the place you are visiting. The localisation in this scenario is not very difficult to achieve, because your smart phone has the assistance from the base stations or even GPS.

In wireless sensor networks, the issue is much more difficult to address. The sensors are low-power devices and intend to be low-cost. This requires hardware and software simplicity, preventing the use of sophisticated hardware or algorithms. Furthermore, the low-power transmission of sensors makes localisation vulnerable to noises. Given these difficulties, localisation is very valuable in sensor networks for the purposes such as routing, sensor deployment and retrieval and security management.

Data Dissemination Using Disruption-Tolerant Networks (DTNs)

contact: Dr Qiang Fu .

When we are sending data over the Internet through desktops, laptops or hand-held devices, we usually assume an end-to-end connectivity can be maintained. However, there are environments where it is difficult or impossible to maintain continuous end-to-end connectivity. Think about the following scenario. In an event of earthquake, network infrastructure is destroyed. A wireless mesh network is quickly deployed. However, we do not have enough network nodes for the required network coverage. We have to move the network nodes around so that a message can be delivered from A to B. If you are lucky, you may be able to maintain instantaneous end-to-end connectivity from A to B. If the routing protocol can establish a route quickly and the MAC protocol can give the bandwidth in time, a packet could be luckily delivered from A to B using an end-to-end approach. However, more likely it is difficult or impossible to maintain end-to-end connectivity. Or, the connectivity can not be maintained long enough for the protocols to act. In this case the end-to-end delivery is impossible. A store-and-forward approach has to be adopted. This project investigates the design issues of network architecture and protocols in disruption/delay tolerant environments.

Effective Spectrum Access in Heterogeneous Wireless Systems

contact: Dr Qiang Fu .

The popularity of wireless communications is driving up the demand on the availability of radio spectrum. Unfortunately, radio spectrum is by nature scarce and expensive. This emphasizes the efficient utilisation of radio spectrum.

However, in practice a large amount of radio spectrum is licensed to particular operators or users. Only licence holders can use the assigned spectrum. This, on one hand, facilitates spectrum management and Quality of Service. But, on the other hand, this also results in inefficient use of spectrum - even if the licensed users are not using the spectrum, it cannot be used by other users. The idea of giving unlicensed users the access to the licensed radio spectrum opens up an opportunity for efficient use of radio spectrum. However, the access has to be made transparent to the licensed users - the licensed users must not be affected when they need to use the spectrum. Unfortunately, this is not a trivial task. There are a few challenges. A user must be able to detect unused spectrum, and determine the spectrum that best suits the requirements of its applications. For instance, some of its applications may require low latency and some may require high throughput. The different requirements may result in selecting different spectrum. A user needs to support seamless mobility, because it is very likely the user has to use spectrum in a dynamic manner. Furthermore, there needs to be a mechanism that can fairly schedule spectrum between users and systems. The challenges are even more evident and real in practice, because of the coexistence of heterogeneous wireless communications systems. A user may have to jump between different systems to maintain or optimise its communication without harmful interference with other users. This project will be focused on formulating these challenges and providing working solutions.

Content Distribution for Vehicular Ad-Hoc Networks (VANETs)

contact: Dr Qiang Fu .

A VANET consists of a dynamic set of vehicles and roadside units. Each of the vehicles and the roadside units can be considered as a network node. A vehicle could have its own Local Area Network (LAN), e.g., a LAN in a train. A roadside unit could be a WiFi base station, a WiMAX base station, a satellite or even a sensor node. A VANET can be a vertical integration of heterogeneous networks. VANETs are considered an example of Mobile Ad-Hoc Networks (MANETs), but have distinct features. VANETs are not as unorganised as the general MANETs - the vehicles move within and along the roads. VANETs are not as infrastructure-less as MANETs, because of the roadside units. These features make VANETs appealing for applications.

Many applications can be developed using VANETs to enhance road safety and passenger comfort. For example, VANETs can be used to deliver short and critical safety messages, real-time multimedia applications or simply download a large amount of data. These applications have different performance requirements: low/high throughput, short/long latency or reliable/unreliable transmission. To accommodate the various performance requirements in such a dynamic and heterogeneous environment is a complex task. This project aims to address the architectural and protocol design principles in VANNETs.