This paper presents a mobility-aware channel allocation strategy for clustered ad hoc network. Our main novelty is to consider the user mobility associated with traffic history and node popularity to guide the channel allocation process, while quickly responding to changes in the network topology. In our performance evaluation and contrarily to related works, we use a realistic mobility model based on user behavior and consider channels with largest spectral distance as well as the smallest number of occurrences. Obtained results show that our strategy presents throughput 15,43 % and 17,74% higher when compared with RANDOM and LD algorithms respectively, and lower overhead when compared with TABU algorithm.