def create_attack(attackedSensor, goal):
usedSensors = (attackedSensor,)
dbOp.connectToDatabase("data/db") ##
# readings & info & also segments
noOfDimensions = dbOp.getNoOfDimensions(rootNodeID = attackedSensor)
nodesSegmentsDic = {}
for nodeID in usedSensors:
readingsInfo = dbOp.selectReadingsFromNode(nodeID)
(dTraining, dTesting, rTr, rTe) = data.getTrainingTesting(readingsInfo)
readings = [Reading(r[0],r[1],r[2],r[3]) for r in rTr]
# get segments
segsInfo = dbOp.selectSegmentsFromNode(nodeID)
segments = [Segment(segInfo[0],segInfo[6],segInfo[1],segInfo[2],segInfo[3],segInfo[4]) for segInfo in segsInfo]
for (i, segment) in enumerate(segments):
segReadings = readings[i*noOfDimensions:(i+1)*noOfDimensions]
segment.set_readings(segReadings)
# set segments
nodesSegmentsDic[nodeID] = segments
# conditional probabilities table
K = dbOp.selectClusterGroup(root_node_id=attackedSensor)[0][1]
condProbsTable = [[0]*K for i in range(K)]
condProbs = dbOp.selectCondProbs(root_node_id=attackedSensor)
for probArray in condProbs:
bCluster = probArray[1]
aCluster = probArray[2]
prob = probArray[3]
condProbsTable[bCluster][aCluster] = prob
dbOp.closeConnectionToDatabase() ##
mcMimicry = MCMimicry()
return (mcMimicry, nodesSegmentsDic, condProbsTable, dTraining, dTesting)
def prep():
#dummy = raw_input('This is gonna take forever (press any key to continue)')
dbOp.connectToDatabase("data/db")
nodes = map(lambda x: x[0], dbOp.selectAllNodes())
for node in nodes:
try:
print ">>", node
readings = dbOp.selectReadingsFromNode(node)
(dTr, dTe, rTr, rTe) = data.getTrainingTesting(readings)
if len(dTr) == 0: raise Exception('No training data')
mcMimicry = MCMimicry(dTr)
(w, segments, centroids, labels, condProbTable, K,
score) = mcMimicry.prepare()
if w is None: continue
# insert cluster group
dbOp.insertClusterGroup(node, K, w)
# insert clusters
for (i, centroid) in enumerate(centroids):
dbOp.insertCluster(node, i, str(centroid))
print ">>>cluster", i
# insert conditional probabilities
for (i, bClusterList) in enumerate(condProbTable):
for (j, aClusterProb) in enumerate(bClusterList):
dbOp.insertConditionalProbability(node, i, j, aClusterProb)
print ">>>", i, j
# insert reading segments
uouo = len(segments)
for (i, segment) in enumerate(segments):
print ">>>segment", i + 1, "/", uouo
start_date = rTr[i * w][1]
start_time = rTr[i * w][2]
end_date = rTr[i * w + w - 1][1]
end_time = rTr[i * w + w - 1][2]
print start_time, '-', end_time
cluster_id = int(labels[i])
dbOp.insertReadingSegment(node, start_date, start_time,
end_date, end_time, node, cluster_id)
dbOp.setNodeAvail(node)
except:
print "Node", node, "failed to initialize"
dbOp.closeConnectionToDatabase()
def prep():
#dummy = raw_input('This is gonna take forever (press any key to continue)')
dbOp.connectToDatabase("data/db")
nodes = map(lambda x: x[0], dbOp.selectAllNodes())
for node in nodes:
try:
print ">>", node
readings = dbOp.selectReadingsFromNode(node)
(dTr, dTe, rTr, rTe) = data.getTrainingTesting(readings)
if len(dTr) == 0: raise Exception('No training data')
mcMimicry = MCMimicry(dTr)
(w, segments, centroids, labels, condProbTable, K, score) = mcMimicry.prepare()
if w is None: continue
# insert cluster group
dbOp.insertClusterGroup(node, K, w)
# insert clusters
for (i, centroid) in enumerate(centroids):
dbOp.insertCluster(node, i, str(centroid))
print ">>>cluster", i
# insert conditional probabilities
for (i, bClusterList) in enumerate(condProbTable):
for (j, aClusterProb) in enumerate(bClusterList):
dbOp.insertConditionalProbability(node, i, j, aClusterProb)
print ">>>", i,j
# insert reading segments
uouo=len(segments)
for (i, segment) in enumerate(segments):
print ">>>segment", i+1,"/",uouo
start_date = rTr[i*w][1]
start_time = rTr[i*w][2]
end_date = rTr[i*w + w -1][1]
end_time = rTr[i*w + w -1][2]
print start_time, '-', end_time
cluster_id = int(labels[i])
dbOp.insertReadingSegment(node, start_date, start_time, end_date, end_time, node, cluster_id)
dbOp.setNodeAvail(node)
except:
print "Node", node, "failed to initialize"
dbOp.closeConnectionToDatabase()
def launch(atck, attackedSensor, usedSensors, goal, tdelay):
dbOp.connectToDatabase("data/db") ##
# readings & info & also segments
noOfDimensions = dbOp.getNoOfDimensions(rootNodeID = attackedSensor)
nodesSegmentsDic = {}
for nodeID in usedSensors:
readingsInfo = dbOp.selectReadingsFromNode(nodeID)
(dTraining, dTesting, rTr, rTe) = data.getTrainingTesting(readingsInfo)
readings = [Reading(r[0],r[1],r[2],r[3]) for r in rTr]
# get segments
segsInfo = dbOp.selectSegmentsFromNode(nodeID)
segments = [Segment(segInfo[0],segInfo[6],segInfo[1],segInfo[2],segInfo[3],segInfo[4]) for segInfo in segsInfo]
for (i, segment) in enumerate(segments):
segReadings = readings[i*noOfDimensions:(i+1)*noOfDimensions]
segment.set_readings(segReadings)
# set segments
nodesSegmentsDic[nodeID] = segments
# conditional probabilities table
K = dbOp.selectClusterGroup(root_node_id=attackedSensor)[0][1]
condProbsTable = [[0]*K for i in range(K)]
condProbs = dbOp.selectCondProbs(root_node_id=attackedSensor)
for probArray in condProbs:
bCluster = probArray[1]
aCluster = probArray[2]
prob = probArray[3]
condProbsTable[bCluster][aCluster] = prob
dbOp.closeConnectionToDatabase() ##
mcMimicry = MCMimicry()
if atck == 0:
(startSignal, iSignal) = mcMimicry.tree_attack(attackedSensor, goal, tdelay, sensorsSegmentsReadingsDic, cond_probs_table)
elif atck == 1:
(startSignal, iSignal) = mcMimicry.greedy_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 2:
(startSignal, iSignal) = mcMimicry.random_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 3:
(startSignal, iSignal) = mcMimicry.greedy_smooth_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 4:
(startSignal, iSignal) = mcMimicry.first_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 5:
(startSignal, iSignal) = mcMimicry.super_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 6:
(startSignal, iSignal) = mcMimicry.rgv_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable, dTesting, propvarderiv=0.65, propdifftemp=0.35, comeBack=1)
elif atck == 7:
(startSignal, iSignal) = mcMimicry.softmax_cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable, dTesting, propvarderiv=0.75, temp=0.03, comeBack=1)
elif atck == 8:
(startSignal, iSignal) = mcMimicry.not_working_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable, dTesting, comeBack=True)
elif atck == 9:
(startSignal, iSignal) = mcMimicry.cluster_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 10:
(startSignal, iSignal) = mcMimicry.ditto_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable, dTesting, comeBack=True)
elif atck == 11:
(startSignal, iSignal) = mcMimicry.rwgm_attack(attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable, dTesting, propMean=0.45, propWDer1=0.2, propDTemp=0.65, comeBack=True)
else:
return None
return (startSignal, iSignal, dTraining, dTesting)
parser.add_argument("--d", nargs='?', default="1", help="Delay in seconds")
args = parser.parse_args()
##########################
# s
##########################
attackedSensor = int(args.s)
usedSensors = (attackedSensor, )
dbOp.connectToDatabase("data/db") ##
# readings & info & segments
noOfDimensions = dbOp.getNoOfDimensions(rootNodeID=attackedSensor)
nodesSegmentsDic = {}
for nodeID in usedSensors:
readingsInfo = dbOp.selectReadingsFromNode(nodeID)
(dTraining, dTesting, rTr, rTe) = data.getTrainingTesting(readingsInfo)
readings = [Reading(r[0], r[1], r[2], r[3]) for r in rTr]
# get segments
segsInfo = dbOp.selectSegmentsFromNode(nodeID)
segments = [
Segment(segInfo[0], segInfo[6], segInfo[1], segInfo[2], segInfo[3],
segInfo[4]) for segInfo in segsInfo
]
for (i, segment) in enumerate(segments):
segReadings = readings[i * noOfDimensions:(i + 1) * noOfDimensions]
segment.set_readings(segReadings)
# set segments
def launch(atck, attackedSensor, usedSensors, goal, tdelay):
dbOp.connectToDatabase("data/db") ##
# readings & info & also segments
noOfDimensions = dbOp.getNoOfDimensions(rootNodeID=attackedSensor)
nodesSegmentsDic = {}
for nodeID in usedSensors:
readingsInfo = dbOp.selectReadingsFromNode(nodeID)
(dTraining, dTesting, rTr, rTe) = data.getTrainingTesting(readingsInfo)
readings = [Reading(r[0], r[1], r[2], r[3]) for r in rTr]
# get segments
segsInfo = dbOp.selectSegmentsFromNode(nodeID)
segments = [
Segment(segInfo[0], segInfo[6], segInfo[1], segInfo[2], segInfo[3],
segInfo[4]) for segInfo in segsInfo
]
for (i, segment) in enumerate(segments):
segReadings = readings[i * noOfDimensions:(i + 1) * noOfDimensions]
segment.set_readings(segReadings)
# set segments
nodesSegmentsDic[nodeID] = segments
# conditional probabilities table
K = dbOp.selectClusterGroup(root_node_id=attackedSensor)[0][1]
condProbsTable = [[0] * K for i in range(K)]
condProbs = dbOp.selectCondProbs(root_node_id=attackedSensor)
for probArray in condProbs:
bCluster = probArray[1]
aCluster = probArray[2]
prob = probArray[3]
condProbsTable[bCluster][aCluster] = prob
dbOp.closeConnectionToDatabase() ##
mcMimicry = MCMimicry()
if atck == 0:
(startSignal,
iSignal) = mcMimicry.tree_attack(attackedSensor, goal, tdelay,
sensorsSegmentsReadingsDic,
cond_probs_table)
elif atck == 1:
(startSignal,
iSignal) = mcMimicry.greedy_attack(attackedSensor, goal, tdelay,
nodesSegmentsDic, condProbsTable)
elif atck == 2:
(startSignal,
iSignal) = mcMimicry.random_cluster_attack(attackedSensor, goal,
tdelay, nodesSegmentsDic,
condProbsTable)
elif atck == 3:
(startSignal, iSignal) = mcMimicry.greedy_smooth_cluster_attack(
attackedSensor, goal, tdelay, nodesSegmentsDic, condProbsTable)
elif atck == 4:
(startSignal,
iSignal) = mcMimicry.first_cluster_attack(attackedSensor, goal,
tdelay, nodesSegmentsDic,
condProbsTable)
elif atck == 5:
(startSignal,
iSignal) = mcMimicry.super_cluster_attack(attackedSensor, goal,
tdelay, nodesSegmentsDic,
condProbsTable)
elif atck == 6:
(startSignal,
iSignal) = mcMimicry.rgv_cluster_attack(attackedSensor,
goal,
tdelay,
nodesSegmentsDic,
condProbsTable,
dTesting,
propvarderiv=0.65,
propdifftemp=0.35,
comeBack=1)
elif atck == 7:
(startSignal,
iSignal) = mcMimicry.softmax_cluster_attack(attackedSensor,
goal,
tdelay,
nodesSegmentsDic,
condProbsTable,
dTesting,
propvarderiv=0.75,
temp=0.03,
comeBack=1)
elif atck == 8:
(startSignal, iSignal) = mcMimicry.not_working_attack(attackedSensor,
goal,
tdelay,
nodesSegmentsDic,
condProbsTable,
dTesting,
comeBack=True)
elif atck == 9:
(startSignal,
iSignal) = mcMimicry.cluster_attack(attackedSensor, goal, tdelay,
nodesSegmentsDic, condProbsTable)
elif atck == 10:
(startSignal, iSignal) = mcMimicry.ditto_attack(attackedSensor,
goal,
tdelay,
nodesSegmentsDic,
condProbsTable,
dTesting,
comeBack=True)
elif atck == 11:
(startSignal, iSignal) = mcMimicry.rwgm_attack(attackedSensor,
goal,
tdelay,
nodesSegmentsDic,
condProbsTable,
dTesting,
propMean=0.45,
propWDer1=0.2,
propDTemp=0.65,
comeBack=True)
else:
return None
return (startSignal, iSignal, dTraining, dTesting)
import dbOp, data
dbOp.connectToDatabase("data/db")
r7 = dbOp.selectReadingsFromNode(7)
dbOp.closeConnectionToDatabase()
(d3Training, d3Testing, r3Tr, r3Te) = data.getTrainingTesting(r7)
import matplotlib.pyplot as plt
#plt.axis('equal')
plt.plot(d3Training, 'b')
plt.show()
parser.add_argument("--d", nargs='?', default="1", help="Delay in seconds")
args = parser.parse_args()
##########################
# s
##########################
attackedSensor = int(args.s)
usedSensors = (attackedSensor,)
dbOp.connectToDatabase("data/db") ##
# readings & info & segments
noOfDimensions = dbOp.getNoOfDimensions(rootNodeID = attackedSensor)
nodesSegmentsDic = {}
for nodeID in usedSensors:
readingsInfo = dbOp.selectReadingsFromNode(nodeID)
(dTraining, dTesting, rTr, rTe) = data.getTrainingTesting(readingsInfo)
readings = [Reading(r[0],r[1],r[2],r[3]) for r in rTr]
# get segments
segsInfo = dbOp.selectSegmentsFromNode(nodeID)
segments = [Segment(segInfo[0],segInfo[6],segInfo[1],segInfo[2],segInfo[3],segInfo[4]) for segInfo in segsInfo]
for (i, segment) in enumerate(segments):
segReadings = readings[i*noOfDimensions:(i+1)*noOfDimensions]
segment.set_readings(segReadings)
# set segments
nodesSegmentsDic[nodeID] = segments
# conditional probabilities table