A Self-Adaptive Optimal Fragmentation Protocol for Multi-Channel Cognitive Radio Ad Hoc Networks

Abstract

In multi-channel cognitive radio ad hoc networks (CRAHNs), packet fragmentation is impacted by new factors besides those in traditional wireless networks due to the unique CR functions. For example, spectrum handoff is the technique for a secondary user (SU) to continue its transmission when a primary user (PU) reoccupies its current transmitting channel. Then, a short frame is less likely to be affected by PU activities, which leads to a lower probability of retransmission. However, with the same header size, a long frame can convey more data than a short frame. In addition, the optimal fragmentation in terms of maximizing the throughput is also related to the spectrum handoff delay, node mobility, and the original packet size of the SU. More importantly, all these factors may vary with time and location, which makes this issue extremely challenging. In this paper, by mathematically modeling these impacts and dynamically mining the related parameters, we propose a self-adaptive protocol guiding the SU to derive the up-to-date optimal packet fragmentation. The proposed protocol is based on practical assumptions and taking other necessary CR functions into account such as spectrum sensing, channel hopping, and spectrum handoff. Simulation results validate our probabilistic model and the optimality of the fragmentation we derived. To the best of our knowledge, this is the first practical fragmentation protocol for multi-channel CRAHNs.