Attacks¶
-
class
graph_tiger.attacks.Attack(graph, runs=10, steps=50, attack='id_node', defense=None, k_d=0, **kwargs)¶ Bases:
graph_tiger.simulations.SimulationThis class simulates a variety of attack strategies on an undirected NetworkX graph
- Parameters
graph – an undirected NetworkX graph
runs – an integer number of times to run the simulation
steps – an integer number of steps to run a single simulation
attack – a string representing the attack strategy to run
defense – a string representing the defense strategy to run
k_d – an integer number of nodes to defend
**kwargs –
see parent class Simulation for additional options
-
reset_simulation()¶ Resets the simulation between each run
-
run_single_sim()¶ Run the attack simulation
-
track_simulation(step)¶ Keeps track of important simulation information at each step of the simulation
- Parameters
step – current simulation iteration
-
graph_tiger.attacks.get_attack_category(method)¶ Gets the attack category e.g., ‘node’, ‘edge’ attack
- Parameters
method – a string representing the attack method
- Returns
a string representing the attack type (‘node’ or ‘edge’)
-
graph_tiger.attacks.get_attack_methods()¶ Gets a list of available attach methods as a list of functions
- Returns
a list of all attack functions
-
graph_tiger.attacks.get_edge_ib(graph, k=3, approx=inf)¶ Get k edges to attack based on Initial Betweenness (IB) Removal [13].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
approx – number of edges to approximate the betweenness centrality
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_id(graph, k=3)¶ Get k edges to attack based on Initial Degree (ID) Removal [13].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_line_deg(graph, k=3)¶ Get k edges to attack using degree centrality by transforming the graph into a line graph [16].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_line_eig(graph, k=3)¶ Get k edges to attack using eigenvector centrality by transforming the graph into a line graph [16].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_line_ns(graph, k=3)¶ Get k edges to attack using Netshield by transforming the graph into a line graph [17,16]
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_line_pr(graph, k=3)¶ Get k edges to attack using PageRank by transforming the graph into a line graph [16].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_rb(graph, k=3, approx=inf)¶ Get k edges to attack based on Recalculated Betweenness (RB) Removal [13].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
approx – number of edges to approximate the betweenness centrality
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_rd(graph, k=3)¶ Get k edges to attack based on Recalculated Degree (RD) Removal [13].
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_edge_rnd(graph, k=3)¶ Randomly select k edges to attack
- Parameters
graph – an undirected NetworkX graph
k – number of edges to attack
- Returns
a list of edge tuples to attack
-
graph_tiger.attacks.get_node_eig(graph, k=3)¶ Get k nodes to attack based on top eigenvector centrality entries
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_ib(graph, k=3, approx=inf)¶ Get k nodes to attack based on Initial Betweenness (IB) Removal [2].
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
approx – number of nodes to approximate the betweenness centrality, k=0.1n is a good approximation, where n
is the number of nodes in the graph
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_id(graph, k=3)¶ Get k nodes to attack based on Initial Degree (ID) Removal [2].
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_ns(graph, k=3)¶ Get k nodes to attack based on the Netshield algorithm :cite`tong2010vulnerability`.
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_pr(graph, k=3)¶ Get k nodes to attack based on top PageRank entries :cite`page1999pagerank`.
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_rb(graph, k=3, approx=inf)¶ Get k nodes to attack based on Recalculated Betweenness (RB) Removal [2].
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
approx – number of nodes to approximate the betweenness centrality, k=0.1n is a good approximation, where n
is the number of nodes in the graph
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_rd(graph, k=3)¶ Get k nodes to attack based on Recalculated Degree (RD) Removal [2].
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.get_node_rnd(graph, k=3)¶ Randomly select k distinct nodes to attack
- Parameters
graph – an undirected NetworkX graph
k – number of nodes to attack
- Returns
a list of nodes to attack
-
graph_tiger.attacks.main()¶
-
graph_tiger.attacks.run_attack_method(graph, method, k=3, approx=None, seed=None)¶ Runs a specified attack on an undirected graph, returning a list of nodes to attack
- Parameters
graph – an undirected NetworkX graph
method – a string representing one of the attack methods
k – number of nodes or edges to attack
approx – attack approximation parameter (not available for every measure)
seed – sets the seed in order to obtain reproducible attacks
- Returns
a list of nodes selected for attack