By IANS,
Washington : A Florida Atlantic University team led by Indian-American Avinash Srinivasan has developed a system that would protect sensitive wireless sensor networks (WSN), engaged in military surveillance, from cyber-attacks.
These WSNs are also extensively used in detection of anti-terrorist activities, hurricanes, earthquakes and forest fires.
The first kind of attack is the fabricated report in which phoney data is sent to the base station with forged validation. Authorities monitoring a WSN for impending disaster are faced with a quandary. If the data arriving from the network is validated but false, how can they know for sure?
Conversely, the second kind of attack seeks to falsify genuine incoming data. How would those monitoring the WSN know whether genuine data is being labelled as false.
Srinivasan and colleagues Feng Li and Jie Wu pointed out that existing WSN systems have built-in software that can ward off the first kind of attack so that false data usually cannot be given valid credentials and those monitoring the system will be able to spot subterfuge easily.
However, WSNs are vulnerable to the second kind of attack, so that a genuine impending disaster cannot be verified remotely, which defeats the purpose of a WSN.
The Florida team has now devised a Probabilistic Voting-based Filtering Scheme (PVFS) to deal with both of these attacks simultaneously. They used a general en-route filtering scheme that can achieve strong protection against hackers while maintaining normal filtering to make the WSN viable.
These findings were published in the International Journal of Security and Networks.
The scheme breaks WSNs into clusters, and locks each cluster to a particular data encryption key. As data reaches headquarters from the WSN clusters, the main cluster-heads along the path checks the report together with the votes, acting as the verification nodes in PVFS.
The verification node is set up so that it will not drop a report immediately it finds a false vote, instead it will simply record the result. Only when the number of verified false votes reaches a designed threshold will a report be dropped.
This way, should a saboteur compromise one or more sensors on any given WSN to launch an attack, the PVFS will apply probability rules to determine the likelihood that this has happened. It will do so based on data arriving from other sensors in different clusters before reporting incoming data as false.
Detecting compromised sensors in a WSN in this way is of vital important to homeland security as well as successfully tracking natural events with the potential to devastate cities.