Selective Repeat - Sliding Window Protocol

The Selective Repeat Protocol (SRP) is a reliable data transmission method that improves upon protocols like Go-Back-N (GBN) by only retransmitting packets that are lost or corrupted. Unlike GBN, which retransmits all packets from the point of failure, SRP is more efficient on unreliable links because it minimizes the number of retransmissions. It requires a full-duplex link, allowing data to be sent in both directions simultaneously.

Key Characteristics

• Window Size: Both the sender's window (Ws​) and the receiver's window (Wr​) are of equal size.
• Packet Handling: The receiver can accept and buffer packets that arrive out of order. This allows it to hold onto correct packets while waiting for a retransmission of a lost one.
• Retransmission: The sender only retransmits specific packets that were not acknowledged after a timeout or upon receiving a Negative Acknowledgment (NAK).
• Efficiency: The efficiency of SRP is given by the formula:

\eta = \frac{N}{1 + 2a}

where N is the window size and a is the ratio of propagation delay to transmission delay (a=Tp​/Tt​).

Why Window Size is Crucial

In Selective Repeat Protocol (SRP), the sender and receiver window size must be ≤ 2^(m−1) (half of the sequence number space). If the window is larger, sequence numbers may repeat before old packets are acknowledged, causing the receiver to confuse new packets with old retransmissions, leading to errors.

Efficiency of Selective Repeat Protocol (SRP) is same as GO-Back-N's efficiency:

\eta = \frac{T_t(\text{data})}{T_t(\text{data}) + 2T_p + T_q + T_{pro} + T_t(\text{ack})}

Where:

\begin{aligned}T_{t,\text{data}} &= \text{Transmission delay of data packet} \\T_p &= \text{Propagation delay} \\T_q &= \text{Queuing delay} \\T_{pro} &= \text{Processing delay} \\T_{t,\text{ack}} &= \text{Transmission delay of acknowledgment}\end{aligned}

Buffers & Sequence Numbers:

• Buffers required = N (sender) + N (receiver)
• Sequence numbers required = N + N = 2N

Efficiency Relation: Same as Go-Back-N

\eta = \frac{1 + 2a}{N}, \quad a = \frac{T_t}{T_p}

Selective Repeat Protocol in Computer Network

Visit Course

Selective Repeat Protocol in Computer Network

Selective Repeat ARQ in Computer Network