Regarding the dynamic nature of cognitive radio networks and the delay of the central decision algorithms due to collecting necessary information in a multi-hop wireless network, distributed resource allocation algorithms are of crucial importance. In this paper we propose a new distributed resource allocation algorithm that properly selects link/frequency in a multi-hop cognitive radio network for delay sensitive applications. This algorithm, uses the locally collected information from neighboring nodes to make suitable routing decisions regarding the dynamic environment and the behavior of the neighbors. We consider the delay as the routing criterion, consisting the queueing delay and the transmission delay. Based on the gathered information from neighboring nodes, we estimate the path delay of all available links and then select the appropriate link/frequency to send data. Simulation results illustrate that our proposed algorithm outperforms the previously published ones that consider only the transmission delay as the decision criterion.