You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

182 lines
6.7 KiB
Python

import mininet
import time, socket, random
import numpy as np
from mininet.cli import CLI
from mininet.log import setLogLevel
from mininet.net import Mininet
from mininet.topo import Topo
from mininet.link import TCLink
import matplotlib.pyplot as plt
# Define custom topology
class MyTopology(Topo):
def build(self):
# Create switches
s1 = self.addSwitch('s1')
s2 = self.addSwitch('s2')
# Create hosts
h1 = self.addHost('h1')
h2 = self.addHost('h2')
# Add links
self.addLink(h1, s1, cls=TCLink, delay='10ms', bw=1)
self.addLink(s1, s2, cls=TCLink, delay='50ms', bw=0.5)
self.addLink(s2, h2, cls=TCLink, delay='10ms', bw=1)
# Define TCP agent
class MyTCPAgent:
def __init__(self):
# Initialize TCP agent
self.transmission_rounds = []
self.congestion_window_sizes = []
def handle_connection(self):
print("TCP connection establishment")
# Implement TCP connection establishment
# Example TCP connection establishment using a socket
# self.sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
# self.sock.connect(('localhost', 19191))
# Example TCP connection establishment simulation
time.sleep(1) # Simulating connection establishment delay
def handle_data_transfer(self, tunnel, window_size):
print("Performing data transfer with tunnel:", tunnel, "and window size:", window_size)
# Implement TCP data transfer with the given tunnel and window size using socket programming
# Implement TCP data transfer simulation
# Placeholder logic: Simulating data transfer
# time.sleep(0.1) # Simulating data transfer delay
# Simulate congestion control by waiting for a fixed amount of time
time.sleep(0.1)
# Record transmission round and final congestion window size
self.transmission_rounds.append(len(self.transmission_rounds) + 1)
self.congestion_window_sizes.append(window_size)
# Define RL agent for tunnel selection
class TunnelSelectionAgent:
def __init__(self, tunnels):
# Initialize your RL agent for tunnel selection
self.rewards = []
self.selected_tunnels = []
self.tunnels = tunnels
self.current_tunnel = 0
def select_tunnel(self):
# Implement tunnel selection based on RL policy
# Placeholder logic: Select a tunnel randomly or based on some criteria
selected_tunnel = self.tunnels[self.current_tunnel]
self.current_tunnel = (self.current_tunnel + 1) % len(self.tunnels)
return selected_tunnel
def update_policy(self, reward):
# Implement RL policy update based on rewards
self.rewards.append(reward)
self.selected_tunnels.append(self.select_tunnel())
# Define RL agent for window prediction
class WindowPredictionAgent:
def __init__(self):
# Initialize the RL agent for window prediction
self.window_sizes = [1, 2, 4, 8, 16, 32, 64] # Possible window sizes
self.alpha = 0.1 # Learning rate
self.gamma = 0.9 # Discount factor
self.q_table = {} # Q-table to store state-action values
def predict_window_size(self):
# Get the current state (e.g., network conditions)
state = self.get_state()
# Check if the state is in the Q-table
if state not in self.q_table:
# Initialize Q-values for all possible actions in the current state
self.q_table[state] = {window_size: 0 for window_size in self.window_sizes}
# Choose the action (window size) based on epsilon-greedy policy
if random.random() < 0.2: # Exploration (20% of the time)
action = random.choice(self.window_sizes)
else: # Exploitation (80% of the time)
action = self.get_best_action(state)
return action
def get_state(self):
# Implement the logic to determine the current state based on network conditions
# For example, you can consider factors such as round-trip time, packet loss rate, or congestion signals
# Placeholder logic: Return a random state
return random.randint(1, 10)
def get_best_action(self, state):
# Find the action (window size) with the highest Q-value for the given state
best_action = max(self.q_table[state], key=self.q_table[state].get)
return best_action
def update_policy(self, reward):
# Update the Q-value based on the reward received after taking an action
# Get the previous state and action
prev_state = self.get_state() # Replace with the actual previous state
prev_action = self.predict_window_size() # Replace with the actual previous action
# Get the current state
curr_state = self.get_state()
# Check if the current state is in the Q-table
if curr_state not in self.q_table:
# Initialize Q-values for all possible actions in the current state
self.q_table[curr_state] = {window_size: 0 for window_size in self.window_sizes}
# Update the Q-value using the Q-learning update rule
max_q_value = max(self.q_table[curr_state].values()) # Get the maximum Q-value for the current state
self.q_table[prev_state][prev_action] += self.alpha * (reward + self.gamma * max_q_value - self.q_table[prev_state][prev_action])
# Main function
if __name__ == '__main__':
setLogLevel('info')
# Create the Mininet network
topo = MyTopology()
net = Mininet(topo=topo)
net.start
# Instantiate TCP and RL agents
tcp_agent = MyTCPAgent()
tunnels = ['myPrivate1']
tunnel_agent = TunnelSelectionAgent(tunnels)
window_agent = WindowPredictionAgent()
# Perform TCP connection establishment
tcp_agent.handle_connection()
# Perform data transfer with RL-based tunnel selection and window prediction
start_time = time.time()
while True:
if time.time() - start_time > 10: # End packet exchange after <X> seconds
break
tunnel = tunnel_agent.select_tunnel()
window_size = window_agent.predict_window_size()
# Perform data transfer
tcp_agent.handle_data_transfer(tunnel, window_size)
# Update RL agents based on rewards
reward = -0.5 # Actual reward value
tunnel_agent.update_policy(reward)
# window_agent.update_policy(reward)
# Stop the Mininet network
net.stop()
# Plot Transmission Round and Congestion Window Size
plt.plot(tcp_agent.transmission_rounds, tcp_agent.congestion_window_sizes)
plt.xlabel('Transmission Round')
plt.ylabel('Congestion Window Size')
plt.title('TCP+RL Window Prediction')
plt.show()