RID: Radio Interference Detection in Wireless Sensor Networks Gang Zhou, Tian He, John A. Stankovic, Tarek Abdelzaher Department of Computer Sceince University of Virginia INFOCOM 2005
Outline Introduction Experiments on radio interference RID Simulation Conclusion
Introduction Interference-connectivity A B C
Introduction B A C
Interference-connectivity Connectivity-interference B A C
Introduction B A C
Experiments on radio interference ~Setup Weak link from T to R 16.2 feet 80% packet delivery ratio Strong link from T to R 8.5 feet 100% packet delivery ratio
Experiments on radio interference ~one direction Radio Interference for a Weak Link Radio Interference for a Strong Link
Experiments on radio interference ~diff. directions Interference Pattern Measured for a Weak Link Interference Pattern Measured for a Strong Link
RID: Radio Interference Detection Protocol HD-ND Detection Information sharing Interference calculation
HD-ND Detection HD packet (with a high sending power) ID information (2 bytes) Packet type (1 byte) To minimize the packet length and to save transmission energy T R HD packet
HD-ND Detection ND packet (with the normal sending power) The ND packet ’ s length is fixed in order that the receiver is able to estimate when the ND packet ’ s transmission will end once it starts to be sensed. T R ND packet
Rules the receiver uses The power level sensed in T1 is as low as that of the background noise On the contrary, receiver thinks this data is useful and records the (transmitter ID,power level) pair for later use Extremely weak, and doesn’t record any information
Add-on rule for receiver to detect disturbance The power level sensed during time period T1, is stable The power level sensed during time period T2, is always as low as that of the background noise.
Add-on rule for receiver to detect disturbance The power level sensed during time period T1, is stable Variable Sensed Power Level During T1
Add-on rule for receiver to detect disturbance The power level sensed during time period T2, is always as low as that of the background noise. Variable Sensed Power Level During T2
Add-on rule for receiver to detect disturbance Stable Sensed Power Level During T1 and T2
Information Sharing T1T1 R T2T2 T3T3 HD ND T1T1 P1P1 Interference_In table
Information Sharing T1T1 RT2T2 T3T3 HDND T1T1 P1P1
Information Sharing T1T1 P1P1 T3T3 P3P3 T1T1 R T2T2 T3T3 HDND T1T1 P1P1
Information Sharing T1T1 P1P1 T3T3 P3P3 APAPA RPRPR T1T1 R T2T2 T3T3 T3T3 P3P3 Interference_In table Interference_Out table Interference_HTP table
Interference Calculation N 2 (D) = {(i 1,i 2 )|(P i1D >receiver_sensitivity) ^ <SNR} i 1 : sender i 2 : jammer D : receiver D i 2 (jammer) i1i1 OK
Interference Calculation N 2 (D) = {(i 1,i 2 )|(P i1D >receiver_sensitivity) ^ < SNR } i 1 : sender i 2 : jammer D : receiver D i 2 (jammer) i1i1 Interference P i1D P i2D +P idle
Interference Calculation The composite of multiple negligible jammers is not necessarily negligible
All collision Scenarios in System N : the number of sensor devices actually deployed {D i } : the set consisting of all nodes in the system D 1 D D k N 2 (D 1 )={ _,_} N 2 (D 2 ) N 2 (D k ) N 3 (D 1 )={ _,_,_} N 3 (D 2 ). { _, _,_}..... N k (D 1 ) N k (D 2) N k (D k )
RID-B : Lightweight Radio Interference Detection Protocol T1T1 P1P1 T3T3 P3P3 T1T1 R T2T2 T3T3
RID-B does not take into consideration the interference cases when multiple transmitters get involved There is no interference calculation phase in RID-B
Simulation GlomoSim
Loss Ratio Simulation #Retransmission
Simulation #Control PacketsTransmission Time
Energy Consumption Simulation
Conclusion Our work is the first to detect radio interference relations among nodes in run- time systems The traditional TDMA protocol, NAMA, can have up to 60% packet loss in heavy load, while the RID-B supported TDMA, NAMA- RID-B, can maintain 100% packet delivery