juneechen/p4_reliable_tcp

CS5700 Computer Networks - Project 4

P4_Reliable_TCP

Project for CS 5700 Computer Networks by Professor Christo Wilson in Spring 2024.

• Contributors: Ujwal Gupta, Shujun Chen

Reliable Communication

This project implements a reliable TCP communication over UDP between a sender and a receiver. The sender sends messages to the receiver, and the receiver acknowledges the receipt of each message. The sender employs congestion control and timeout-based retransmission mechanisms to ensure reliable delivery of messages.

High-Level Approach

• Sender:

  • Utilize UDP sockets for communication.
  • Attach checksum to packets.
  • Maintain a seq_num and last_ack_seq to keep track of packet status.
  • Keep a copy of all in-flight packets with meta info including sent time, timeout time, re-transmit or not, etc.
  • Handle accumulative ack and remove packets from in_flight list properly.
  • Implements congestion control using a sliding window mechanism.
  • Estimate the RTT with each sample RTT from all non-retranmitted packets.
  • Maintain a timeout value, which is dynamictic adjusted by estimating RTT.
  • Dynamically adjust cwnd based on acknowledgments and timeouts, in response to various network enviroments and incidents.
  • Simulate TCP Reno's behavior of fast-retransmission.
  • Handles acknowledgments, timeouts, and retransmissions reliably.

• Receiver:

  • Binds to a UDP socket to receive messages.
  • Verifies checksums for received packets to detect corruption.
  • Sends cumulative acknowledgments to the sender.
  • Maintain a next_expected sequence number.
  • Hold on to all out of order packets in an recved dictionary
  • Reconstructs the original message from received packets.

Challenges Faced

• Reliability Over UDP: Implementing reliability over UDP required handling acknowledgments, timeouts, and retransmissions to ensure message delivery.
• Congestion Control: Designing an efficient congestion control mechanism to regulate the flow of data while maximizing throughput without causing congestion collapse.
• Checksum Calculation: Ensuring accurate checksum calculation to detect corrupted packets while minimizing computational overhead.

Key Features

• Reliability: Implements acknowledgment-based reliability mechanisms to ensure message delivery.
• Congestion Control: Utilizes a sliding window and congestion window to regulate the flow of data.
• Checksum Verification: Computes and verifies checksums for detecting corrupted packets.
• Timeout-Based Retransmissions: Retransmits packets upon timeout to handle packet loss.
• Cumulative Acknowledgments: Sends cumulative acknowledgments to acknowledge received packets efficiently.

Testing Approach

• All tests done using the testing network simulator.
• Unit Testing: Conducted extensive unit testing for individual components and functionalities.
• Integration Testing: Tested the interaction between sender and receiver to ensure seamless communication.
• Stress Testing: Simulated various network conditions and loads to evaluate the robustness and scalability of the protocol.
• Error Handling Testing: Tested error handling mechanisms for scenarios such as packet corruption, timeout, and unexpected behavior.