Introducing Epidemic Models for Data Survivability in UWSNs Roberto Di Pietro, Nino Vincenzo Verde {dipietro,nverde}@mat.uniroma3.it Universita’ di Roma Tre RoadMap • UWSNs • Epidemic Models – SIR – SIS • Epidemic Models for Information Survivability in UWSNs – Modeling the problem – Applications – Results • Conclusions Unattended WSNs • Sporadic presence of the sink • Sensors upload info as soon as the sink comes around • Applications: – Hostile environment monitoring – Pipeline monitoring Information Survivability • Sink not always available: – More subject to malicious attacks than traditional WSN • Our target: To provide a certain level of assurance about Information Survivability Epidemic Models • Stochastic approaches – Accurately describe fluctuation – Variation chance in exposure risk – Very complex – Laborious to set up Epidemic Models • Deterministic approaches – Describe the dynamic of a disease at the population scale – Fits very large populations – Not accurate with small populations Are they accurate when trying to minimize “energy consumption”? Deterministic Epidemic Models • n individuals are partitioned into several compartment • Transition probabilities between any two compartments are given • The spreading of the disease is taken into consideration C2 C1 C3 SIR S SI I Infected Susceptibles I R Recovered s ' (t ) s (t )i (t ) i ' (t ) s (t )i (t ) i (t ) r ' (t ) i (t ) • It does not admit a generic analytic solution, but: s(t ) s(0)e ( r ( t ) r ( 0 )) Is the basic reproduction number: if >= 1/s(0) then Epidemic outbreak SIS SI S Susceptibles I I Infected i ' (t ) s (t )i (t ) i (t ) s ' (t ) i (t ) s (t )i (t ) • Solution: i (t ) ( ) e t ( ) c ( ) • Using i(t) it is possible to predict the number of sick individuals at time t Modeling the Information Spread • n sensors, 1 sink, 1 attacker • A secure routing protocol allows to exchange information between any pair of sensor • Evolution time partitioned in rounds – Both sensors and the attackers play their game s(t ) is t hefract ionof sensors t hatdo not possess t hedat um i (t ) is t hefract ionof sensorspossesingit r (t ) is t hefract ionof sensorsdest royedby t heat t acker Sensors Model • Data is transmitted by replication: – Each sensor that stores the datum transmits it with probability to each neighbor n • Theorem: If i is the fraction of sensors possessing the datum, and if each sensor forwards the datum with probability α/n , the value siα is an approximation of the probability that the datum reaches a sensor that do not possess it. Proof Probability that no sensors sent the datum to Na in s 1 1 n Sensors that currently possess the datum siα Probability that Na did not possess the datum Probability that at least one sensor sent the datum to Na α/n is close to 0 -> using the binomial approximation the above formula is equal to αsi Attacker Model • We consider 2 attackers: – ADVsimple: • It is able to destroy each sensors containing the datum with probability β in each time step – ADVstealth: • It is able to erase the datum without destroying the sensor • It does not change the behavior of the sensor Epidemic models in UWSNs ADVsimple Replication α/n SIR ADVstealth Replication α/n SIS SIR Test • • • • α=0.605 β=0.5 n=100 i(0)=0.1 SIS Test • n=100 • i(0)=0.1 Outcome • Epidemic models can be used to forecast the behavior of an UWSNs • It becomes easy to set-up the parameters – It is possible to study the conditions that have to be satisfied to assure information survivability • Problems? – Energy Consumption: it is needed to minimize the replication process Minimizing Energy Consumption • In both the models energy consumption is minimized when i(t) is close to 0 for any t • Statistical fluctuation can force the system to loose the datum n=100; i(0)=0.05 n=100; i(0)=0.01 Video Simulation SIS n=100; α=0.22; β=0.2 Steady state when i(t)=(1-β/α) Is the information survivability assured? Conclusions • Deterministic epidemic models can be used to model information assurance in UWSNs – The parameters that assure the survivability are easy to set up – They fit very well large sensor networks • Unlikely events can induce the loss of the datum – It is needed to assess bounds on the probability of these events Questions? Thank you! Some Related Work • R. Di Pietro, and N. V. Verde. Epidemic data survivability in Unattended Wireless Sensor Networks. In Proceedings of the ACM Conference on Wireless Network Security (WiSec), Hamburg, Germany, June 2011. • Michele Albano, Stefano Chessa, and Roberto Di Pietro. “A model with applications for data survivability in Critical Infrastructures”. In Journal of Information Assurance and Security, vol. 4(6), pages 629-639, June 2009. • Roberto Di Pietro, Luigi V. Mancini, Claudio Soriente, Angelo Spognardi, and Gene Tsudik. “Catch Me (If You Can): Data Survival in Unattended Sensor Networks”. In Proceedings of the 6th IEEE International Conference on Pervasive Computing and Communications (PerCom 2008), pages 185-194, Hong Kong, March 17-21, 2008. • Roberto Di Pietro, Luigi V. Mancini, Claudio Soriente, Angelo Spognardi, and Gene Tsudik. “Playing Hide-and-Seek with a Focused Mobile Adversary in Unattended Wireless Sensor Networks”. In Journal of Ad Hoc Networks (Elsevier) - Special Issue on Privacy and Security in Wireless Sensor and Ad Hoc Networks -, vol. 7(8), pages 1463-1475, November 2009.