Mobile Computing | DAVV Unit 4

Overview of TCP/IP:

TCP/IP is a set of networking protocols used for communication over the internet and local networks. It allows computers and devices to share data in a standardized way.

Architecture of TCP/IP:

TCP/IP is organized into a four-layer architecture, each serving a specific purpose:

1. Application Layer: This top layer contains protocols that directly interact with user applications. Examples include HTTP (web browsing), SMTP (email), FTP (file transfer), and DNS (domain name resolution).
2. Transport Layer: The transport layer manages end-to-end communication and data flow control. Two main protocols at this layer are TCP (Transmission Control Protocol) and UDP (User Datagram Protocol).
3. Internet Layer: Responsible for addressing, routing, and fragmentation of data packets. The Internet Protocol (IP) is the core protocol of this layer. It ensures that data packets reach their destination through the most efficient path.
4. Link Layer: This layer handles the physical connection to the network medium, such as Ethernet or Wi-Fi. It manages the transmission of data frames over the physical medium.

Adaptation of TCP Window:

The TCP window, also called the sliding window, is an important part of how TCP controls the flow of data. It tells the sender how much data it can send before needing confirmation from the receiver.

Adaptive Window: TCP adapts the window size dynamically based on the network conditions. If the network is stable and congestion-free, the window size may increase to utilize available bandwidth effectively. If congestion or packet loss occurs, the window size may decrease to prevent further congestion.

Improvement in TCP Performance:

TCP has undergone several optimizations to enhance its performance in various network environments:

1. TCP Congestion Control: TCP employs algorithms to detect and respond to network congestion. Algorithms like TCP Reno, NewReno, and TCP Cubic manage network congestion to prevent excessive packet loss and ensure fair resource utilization.
2. TCP Fast Retransmit and Recovery: These mechanisms enable TCP to recover from packet loss quickly without needing to wait for a full timeout, thereby reducing latency from retransmissions.
3. Selective Acknowledgment (SACK): SACK enables the receiver to inform the sender about received packets, allowing the sender to retransmit only missing segments, minimizing unnecessary retransmissions.
4. TCP Offloading: Offloading TCP processing to network hardware or specialized co-processors improves performance by reducing the CPU load and increasing throughput.
5. TCP Segmentation Offload (TSO) and Large Receive Offload (LRO): These techniques optimize data segment handling, reducing overhead for processing smaller segments.
6. Multipath TCP (MPTCP): MPTCP enables a single TCP connection to use multiple paths simultaneously, enhancing throughput and reliability, particularly in scenarios with multiple network interfaces.