Goodput Analysis for Burst-Noisy Low-Speed Wireless Networks During Message Segmentation

This paper derives the analytical form of goodput for wireless networks where packets generated from a message through a segmentation function are lost due to burst bit errors, and these lost packets are recovered by a stop-and-wait protocol. In previous works, the message segmentation behavior has not been taken into account. To solve this problem, we propose an approximate analytical model using the exact distribution of packet information field sizes (PIFSs), given a message size distribution and payload size. For a simple scenario where fixed-sized messages are transferred using IPv6 over a LoWPAN, we show that: 1) a larger mean bit error burst length yields larger goodput values under the same mean bit error rate because packet corruption probabilities are smaller, 2) the traditional approximation where all PIFSs are equal to the payload size yields optimistic values, and 3) the optimal payload size derived under the assumption of independently occurring bit errors is noticeably smaller compared with when bit errors occur in bursts.