Performance evaluation of network coding in IEEE 802.11 wireless ad hoc networks

We evaluate the practical network coding (NC) gain in a wireless ad hoc networks. First, we introduce how network coding can be applied to IEEE 802.11 ad hoc networks. Next, we find obstructive factors by which the NC gain may decrease as compared with the theoretical NC gain. Finally, through the performance evaluation, we analyze why a performance difference occurs between practical and theoretical NC gains owing to the suggested obstructive factors. According to the simulated results, even though there is some overhead caused by the characteristics of the media access control (MAC) protocol and control signals to apply NC to wireless ad hoc networks, the practical NC gain is almost the same as the theoretical gain in an ideal environment for NC. However, in the NC system with the hidden node problem, the NC gain decreases because of packet collision. In the NC system using a promiscuous mode where packet overhearing is possible, nodes consume more power than the conventional system because of overhearing packets. For instance, when the number of overhearing nodes not related to NC is set to 15, the power reduction rate is -38.04%. Furthermore, under the network conditions inappropriate for NC, such as asymmetric flows and scarce coding structures, the NC gain can be considerably reduced. For example, in 5x5 and 7x7 grid topology networks, the load reduction rate by NC are only 4.12% and 7.01%, respectively. Therefore, research to solve the suggested obstructive factors is necessary to achieve sufficient NC gain in a practical wireless network such as an IEEE 802.11 ad hoc networks.