Andrea Passarella - Research Topics

Here you can find a summary on my research activities. Take a look at the publications’ page (or mail me) for more details.

Opportunistic Networking

Opportunistic Networking (aka Delay-Tolerant Networking) is one of the most interesting evolution of the ad hoc networking paradigm.

In legacy ad hoc networks nodes of a MANET are supposed to be well connected to each other, in the sense that there has to exist a (multi-hop) path connecting a source and a destination node willing to communicate. If the nodes are not connected at the same time, messages get dropped at some point in the network. Opportunistic Networks release this constraint, and enable nodes that are not necessarily connected at the same time to communicate nevertheless.

The main topics we are investigating within the Haggle and Socialnets projects are:

• content management

• routing/forwarding protocols

• mobility models

The Opportunistic Networks concept implies a number o new challenges. Messages have not to be discarded when eventual destination are disconnected, but should be temporary buffered at some node. Legacy routing protocols designed for MANETs have to be drastically re-designed. Most importantly, next hops have to be chosen dynamically, based on the current context available at forwarding nodes, instead of being computed based only on topological information. Data should be disseminated and replicated to increase availability. Reseource management (e.g., memory, energy, bandwidth) has to be wisely managed, as data replication and forwarding policies might require higher resource usage than in legacy MANETs.

Wireless Sensor Networks

WASN are one of the hottest topics in the mobile computing field. Within the EC Memory project we are investigating models of the humans visual spatio-temporal perception (specifically, in cases of saccadic distortions). The understanding of these phenomena will be exploited to design WSN systems that could exploit similar effects, or could anticipate them, in order to prevent distortions.

Before working on Memory, I worked on experimental measurements, and on energy-saving issues. Unfortunately, most of the works on WSN are not based on experimental evidences, and just relies on (sometime quite questionable) analytical and simulation models. We contributed to fill the gap, by caracterising trough experiments the main networking features of Mica MOTES, the leading technology for sensor networks.

This activity was carried out as follows:

• Experimental Activity: we have run extensive measurements on Mica Mote networks. We have characterised Mica Motes with respect to the energy consumption and the networking capabilities (e.g., transmission and carrier sensing range). Based on these measurements, we have defined a channel model for Mica Mote networks.

• Design Activity: in collaboration with the Rutgers University, we have designed new power-saving architecture and protocols for wireless sensor networks.

• Testing Activity: we have carried out a preliminary analysis of the power-saving protocols by means of analytical models.

• Reference Projects: VICOM, DataSpace.

P2P for Mobile Ad hoc Networks

Group-Communication Applications are an outstanding opportunity to bring MANETs in the real world. Their common requirements, and the self-organising nature of MANETs, suggest to support them through middleware-level P2P systems. Standard P2P systems designed for wired networks generate too much overhead in MANETs, and hence optimised P2P implementations have to be devised.

This activity (part of the MobileMAN project) was carried on as follows:

• Design Activity: we have designed and implemented a P2P Group-Communication Application for Ad hoc Networks (The Whiteboard). The Whiteboard and its p2p support have been implemented in prototype ad hoc network stacks. Currently, we are designing p2p multicast protocols optimized for ad hoc networks by leveraging the cross-layer paradigm.

• Testing Activity: we are testing the ad hoc network prototypes through real experiments. We characterise the Whiteboard performance with respect to the user QoS, and we also characterise the quality of the multicast tree used to deliver data to application users.

Power-saving Architectures and Protocols for wireless hotspots

WLAN installations are ever more widespread. Unfortunately, you still have to be close to a power socket to enjoy your wireless-Internet access, since WLAN card consume a lot of energy! Most of my Ph.D. has been spent on looking for techniques for reducing the WLAN card energy consumption without affecting the QoS perceived by the user.

This activity was carried out as follows:

• Design Activity: we have designed an architecture based on the Indirect-TCP model for power-saving mobile Internet access. Within this architecture, several power-saving policies have been defined, along with the respective network protocols. A Cross-layer Power Manager has been designed, integrating these policies and the standard 802.11 Power-Saving Mode.

• Testing Activity: we have implemented the new power-saving network protocols in a real-Internet prototype. We have evaluated the effectiveness of our architecture with respect to several traffic patterns. We also used the measurements from the prototype to validate analytical models of the protocols’ behavior. The effectiveness of integrating the new power-saving policies and the 802.11 standard has been evaluated by means of analytical and simulation models.

• Reference Projects: “Web Systems with QoS guarantees”, “Internet: Efficiency, Integration and Security”, and “VICOM (Virtual Immersive COMmunications)”.

Transport Protocols for Ad hoc Networks

The legacy TCP protocol is known to perform poorly in MANETs. Many features contribute to its far-from-optimal behavior. Instead of widely patching TCP, we believe it is more fruiteful to design a completely new protocol, and postponing compatibility issues at a later stage.

This activity was carried out as follows:

• Design Activity: we have designed a novel transport protocol for ad hoc networks (TPA). We have redesigned the functionalities that make TCP not suited to work in ad hoc environments.

• Testing Activity: we are currently testing the protocol by means of simulation models.

• Reference Projects: VICOM, MobileMAN.