README.md

Simulation of the Qasino model

It uses the PeerSim P2P simulator.

Requirements :

• Apache Maven 3.6.0
• Java 8 or Openjdk 11
• Python 3.7.3

mvn clean package

Intermediate Results of all queries of the paper

	TP1	TP2	TP3	TP4	TP5	TP6	TP7	Total
q17	1	1						2
q22	1	4213						4214
q54	1	2889	1284	1284				5458
q73	2	4213	2889	9670	1284	1284		19342
q87	1	1	1	1	1	1	4	10

Las Vegas

Number of iterations of the LV algorithms in order to get all nodes for different network size, against the approximation.

Monte Carlo (one node execute a query)

Different number of nodes yet to be seen after an execution of the MC algorithm for different value of p (where p is the wanted proportion of nodes to reach)

The network size is set to 1000

Monte Carlo (q nodes execute a query) for N = 1000 and p = 0.97

For all queries and a total of 100 runs,

• left: this is the number of iterations needed for terminating the execution of the MC algorithm with different number of nodes executing the queries in the network and a p fixed to 0.97
• right: the proportion of completed executions for all runs in percentage

All results are available for each query in results/plots/p-0.97-q*.png

The traffic in triples received using Invertible bloom lookup tables for each query

An IBLT is set to perform a set difference over a maximum of 500 differences and alpha is set to 2 So an IBLT has a size of 1000 cells of 3 32-bits integers (keysum, hashsum, count). So an IBLT has a size of 32k bytes when sending over the network.

Query 17

Query 22

Query 54

Query 73

Query 87

Network size estimator

The original estimator is: Jelasity, M., & Montresor, A. (2004). Epidemic-style proactive aggregation in large overlay networks. In 24th International Conference on Distributed Computing Systems, 2004. Proceedings. (pp. 102-109). IEEE.

Network size estimation for 1000 cycles over a network of 1000 peers.

Number of instances computing the estimation per peer during the experiments