import parser.parse_dns
+# List of devices
DEVICE_MAC_LIST = "devicelist.dat"
+EXCLUSION_MAC_LIST = "exclusion.dat"
COLUMN_MAC = "MAC_address"
COLUMN_DEVICE_NAME = "device_name"
-
-
+# Fields
JSON_KEY_SOURCE = "_source"
JSON_KEY_LAYERS = "layers"
JSON_KEY_FRAME = "frame"
+JSON_KEY_FRAME_PROTOCOLS = "frame.protocols"
JSON_KEY_FRAME_TIME_EPOCH = "frame.time_epoch"
+JSON_KEY_FRAME_LENGTH = "frame.len"
JSON_KEY_ETH = "eth"
JSON_KEY_ETH_SRC = "eth.src"
JSON_KEY_ETH_DST = "eth.dst"
JSON_KEY_UDP = "udp"
JSON_KEY_TCP = "tcp"
# List of checked protocols
-listchkprot = [ "bootp",
+listchkprot = [ "arp",
+ "bootp",
"dhcpv6",
"dns",
"llmnr",
"ssdp" ]
-def parse_json(file_path):
-
+def create_device_list(dev_list_file):
+ """ Create list for smart home devices from a CSV file
+ Args:
+ dev_list_file: CSV file path that contains list of device MAC addresses
+ """
# Open the device MAC list file
- with open(DEVICE_MAC_LIST) as csvfile:
- maclist = csv.DictReader(csvfile, (COLUMN_MAC, COLUMN_DEVICE_NAME))
- crudelist = list()
- for item in maclist:
- crudelist.append(item)
- #print(item)
+ with open(dev_list_file) as csvfile:
+ mac_list = csv.DictReader(csvfile, (COLUMN_MAC, COLUMN_DEVICE_NAME))
+ crude_list = list()
+ for item in mac_list:
+ crude_list.append(item)
# Create key-value dictionary
- devlist = dict()
- for item in crudelist:
- devlist[item[COLUMN_MAC]] = item[COLUMN_DEVICE_NAME]
+ dev_list = dict()
+ for item in crude_list:
+ dev_list[item[COLUMN_MAC]] = item[COLUMN_DEVICE_NAME]
#print item["MAC_address"] + " => " + item["device_name"]
#for key, value in devlist.iteritems():
# print key + " => " + value
+ return dev_list
+
+
+def traverse_and_merge_nodes(G, dev_list_file):
+ """ Merge nodes that have similar properties, e.g. same protocols
+ But, we only do this for leaves (outer nodes), and not for
+ nodes that are in the middle/have many neighbors.
+ The pre-condition is that the node:
+ (1) only has one neighbor, and
+ (2) not a smarthome device.
+ then we compare the edges, whether they use the same protocols
+ or not. If yes, then we collapse that node and we attach
+ it to the very first node that uses that set of protocols.
+ Args:
+ G: a complete networkx graph
+ dev_list_file: CSV file path that contains list of device MAC addresses
+ """
+ nodes = G.nodes()
+ #print "Nodes: ", nodes
+ node_to_merge = dict()
+ # Create list of smarthome devices
+ dev_list = create_device_list(DEVICE_MAC_LIST)
+ # Traverse every node
+ # Check that the node is not a smarthome device
+ for node in nodes:
+ neighbors = G.neighbors(node)
+ #print "Neighbors: ", neighbors, "\n"
+ # Skip if the node is a smarthome device
+ if node in dev_list:
+ continue
+ # Skip if the node has many neighbors (non-leaf) or no neighbor at all
+ if len(neighbors) is not 1:
+ continue
+ #print "Node: ", node
+ neighbor = neighbors[0]
+ #print "Neighbor: ", neighbors
+ protocols = G[node][neighbor]['Protocol']
+ #print "Protocol: ", protocols
+ # Store neighbor-protocol as key in dictionary
+ neigh_proto = neighbor + "-" + protocols
+ if neigh_proto not in node_to_merge:
+ node_to_merge[neigh_proto] = node
+ else:
+ # Merge this node if there is already an entry
+ # First delete
+ G.remove_node(node)
+ node_to_merge_with = node_to_merge[neigh_proto]
+ merged_nodes = G.node[node_to_merge_with]['Merged']
+ # Check if this is the first node
+ if merged_nodes is '':
+ merged_nodes = node
+ else:
+ # Put comma if there is already one or more nodes
+ merged_nodes += ", " + node
+ # Then attach as attribute
+ G.node[node_to_merge_with]['Merged'] = merged_nodes
+
+ return G
+
+
+def place_in_graph(G, eth_src, eth_dst, device_dns_mappings, dev_list, layers,
+ edge_to_prot, edge_to_vol):
+ """ Place nodes and edges on the graph
+ Args:
+ G: the complete graph
+ eth_src: MAC address of source
+ eth_dst: MAC address of destination
+ device_dns_mappings: device to DNS mappings (data structure)
+ dev_list: list of existing smarthome devices
+ layers: layers of JSON file structure
+ edge_to_prot: edge to protocols mappings
+ edge_to_vol: edge to traffic volume mappings
+ """
+ # Get timestamp of packet (router's timestamp)
+ timestamp = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_TIME_EPOCH])
+ # Get packet length
+ packet_len = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_LENGTH])
+ # Get the protocol and strip just the name of it
+ long_protocol = layers[JSON_KEY_FRAME][JSON_KEY_FRAME_PROTOCOLS]
+ # Split once starting from the end of the string and get it
+ split_protocol = long_protocol.split(':')
+ protocol = None
+ if len(split_protocol) < 5:
+ last_index = len(split_protocol) - 1
+ protocol = split_protocol[last_index]
+ else:
+ protocol = split_protocol[3] + ":" + split_protocol[4]
+ print "timestamp: ", timestamp, " - new protocol added: ", protocol, "\n"
+ # Store protocol into the set (source)
+ protocols = None
+ # Key to search in the dictionary is <src-mac-address>-<dst-mac_address>
+ dict_key = eth_src + "-" + eth_dst
+ if dict_key not in edge_to_prot:
+ edge_to_prot[dict_key] = set()
+ protocols = edge_to_prot[dict_key]
+ protocols.add(protocol)
+ protocols_str = ', '.join(protocols)
+ print "protocols: ", protocols_str, "\n"
+ # Check packet length and accumulate to get traffic volume
+ if dict_key not in edge_to_vol:
+ edge_to_vol[dict_key] = 0;
+ edge_to_vol[dict_key] = edge_to_vol[dict_key] + packet_len
+ volume = str(edge_to_vol[dict_key])
+ # And source and destination IPs
+ ip_src = layers[JSON_KEY_IP][JSON_KEY_IP_SRC]
+ ip_dst = layers[JSON_KEY_IP][JSON_KEY_IP_DST]
+ # Categorize source and destination IP addresses: local vs. non-local
+ ip_re = re.compile(r'\b192.168.[0-9.]+')
+ src_is_local = ip_re.search(ip_src)
+ dst_is_local = ip_re.search(ip_dst)
+ print "ip.src =", ip_src, "ip.dst =", ip_dst, "\n"
+ # Place nodes and edges
+ src_node = None
+ dst_node = None
+ if src_is_local:
+ G.add_node(eth_src, Name=dev_list[eth_src])
+ src_node = eth_src
+ else:
+ hostname = None
+ # Check first if the key (eth_dst) exists in the dictionary
+ if eth_dst in device_dns_mappings:
+ # If the source is not local, then it's inbound traffic, and hence the eth_dst is the MAC of the IoT device.
+ hostname = device_dns_mappings[eth_dst].hostname_for_ip_at_time(ip_src, timestamp)
+ if hostname is None:
+ # Use IP if no hostname mapping
+ hostname = ip_src
+ # Non-smarthome devices can be merged later
+ G.add_node(hostname, Merged='')
+ src_node = hostname
+
+ if dst_is_local:
+ G.add_node(eth_dst, Name=dev_list[eth_dst])
+ dst_node = eth_dst
+ else:
+ hostname = None
+ # Check first if the key (eth_dst) exists in the dictionary
+ if eth_src in device_dns_mappings:
+ # If the destination is not local, then it's outbound traffic, and hence the eth_src is the MAC of the IoT device.
+ hostname = device_dns_mappings[eth_src].hostname_for_ip_at_time(ip_dst, timestamp)
+ if hostname is None:
+ # Use IP if no hostname mapping
+ hostname = ip_dst
+ # Non-smarthome devices can be merged later
+ G.add_node(hostname, Merged='')
+ dst_node = hostname
+ G.add_edge(src_node, dst_node, Protocol=protocols_str, Volume=volume)
+
+
+def parse_json(file_path):
+ """ Parse JSON file and create graph
+ Args:
+ file_path: path to the JSON file
+ """
+ # Create a smart home device list
+ dev_list = create_device_list(DEVICE_MAC_LIST)
+ # Create an exclusion list
+ exc_list = create_device_list(EXCLUSION_MAC_LIST)
# First parse the file once, constructing a map that contains information about individual devices' DNS resolutions.
device_dns_mappings = parser.parse_dns.parse_json_dns(file_path) # "./json/eth1.dump.json"
-
# Init empty graph
- G = nx.DiGraph()
+ G = nx.DiGraph()
+ # Mapping from edge to a set of protocols
+ edge_to_prot = dict()
+ # Mapping from edge to traffic volume
+ edge_to_vol = dict()
# Parse file again, this time constructing a graph of device<->server and device<->device communication.
with open(file_path) as jf:
- # Read JSON.
- # data becomes reference to root JSON object (or in our case json array)
+ # Read JSON; data becomes reference to root JSON object (or in our case json array)
data = json.load(jf)
-
# Loop through json objects (packets) in data
for p in data:
- # p is a JSON object, not an index
- # Drill down to object containing data from the different layers
+ # p is a JSON object, not an index - drill down to object containing data from the different layers
layers = p[JSON_KEY_SOURCE][JSON_KEY_LAYERS]
iscontinue = False
if JSON_KEY_UDP not in layers and JSON_KEY_TCP not in layers:
continue
- # Fetch timestamp of packet (router's timestamp)
- packet_timestamp = Decimal(layers[JSON_KEY_FRAME][JSON_KEY_FRAME_TIME_EPOCH])
- print "timestamp", packet_timestamp
# Fetch source and destination MACs
eth = layers.get(JSON_KEY_ETH, None)
if eth is None:
continue
eth_src = eth.get(JSON_KEY_ETH_SRC, None)
eth_dst = eth.get(JSON_KEY_ETH_DST, None)
- # And source and destination IPs
- ip_src = layers[JSON_KEY_IP][JSON_KEY_IP_SRC]
- ip_dst = layers[JSON_KEY_IP][JSON_KEY_IP_DST]
-
- # Categorize source and destination IP addresses: local vs. non-local
- ipre = re.compile(r'\b192.168.[0-9.]+')
- src_is_local = ipre.search(ip_src)
- dst_is_local = ipre.search(ip_dst)
- print "ip.src =", ip_src, "ip.dst =", ip_dst
-
- src_node = None
- dst_node = None
- if src_is_local:
- G.add_node(eth_src, Name=devlist[eth_src])
- src_node = eth_src
- else:
- hostname = None
- # Check first if the key (eth_dst) exists in the dictionary
- if eth_dst in device_dns_mappings:
- # If the source is not local, then it's inbound traffic, and hence the eth_dst is the MAC of the IoT device.
- hostname = device_dns_mappings[eth_dst].hostname_for_ip_at_time(ip_src, packet_timestamp)
- if hostname is None:
- # Use IP if no hostname mapping
- hostname = ip_src
- G.add_node(hostname)
- src_node = hostname
- if dst_is_local:
- G.add_node(eth_dst, Name=devlist[eth_src])
- dst_node = eth_dst
- else:
- hostname = None
- # Check first if the key (eth_dst) exists in the dictionary
- if eth_src in device_dns_mappings:
- # If the destination is not local, then it's outbound traffic, and hence the eth_src is the MAC of the IoT device.
- hostname = device_dns_mappings[eth_src].hostname_for_ip_at_time(ip_dst, packet_timestamp)
- if hostname is None:
- # Use IP if no hostname mapping
- hostname = ip_dst
- G.add_node(hostname)
- dst_node = hostname
- G.add_edge(src_node, dst_node)
+ # Exclude devices in the exclusion list
+ if eth_src in exc_list:
+ print "[ WARNING: Source ", eth_src, " is excluded from graph! ]"
+ continue
+ if eth_dst in exc_list:
+ print "[ WARNING: Destination ", eth_dst, " is excluded from graph! ]"
+ continue
+
+ # Place nodes and edges in graph
+ place_in_graph(G, eth_src, eth_dst, device_dns_mappings, dev_list, layers,
+ edge_to_prot, edge_to_vol)
# Print DNS mapping for reference
- for mac in device_dns_mappings:
- ddm = device_dns_mappings[mac]
- ddm.print_mappings()
+ #for mac in device_dns_mappings:
+ # ddm = device_dns_mappings[mac]
+ # ddm.print_mappings()
return G
+
# ------------------------------------------------------
# Not currently used.
# Might be useful later on if we wish to resolve IPs.
return False
# ------------------------------------------------------
+
if __name__ == '__main__':
if len(sys.argv) < 3:
print "Usage:", sys.argv[0], "input_file output_file"
print "[ output_file =", output_file, "]"
# Construct graph from JSON
G = parse_json(input_file)
+ # Contract nodes that have the same properties, i.e. same protocols
+ G = traverse_and_merge_nodes(G, DEVICE_MAC_LIST)
# Write Graph in Graph Exchange XML format
nx.write_gexf(G, output_file)
projects / pingpong.git / blobdiff