Abstract: Ant colony algorithm is easy to fall in local best and its convergent speed is slow in solving multiple QoS constrained unicast routing problems.Therefore，a differential pheromone ant colony algorithm is proposed to solve the problems.Iterators and the function of adjusting differential pheromone intensity on the path are used to adjust nodes transition strategy dynamically.Stagnation behavior is judged by the concentration level of the pheromone on the path.Path mutated and twice ant colony algorithm search are adopted and the pheromone is limited according to the principles of MAX-MIN ant system.According to simulations，its global search is strong and it can range out of local best and fast converge to the global optimum.The improved algorithm is feasible and effective.

Key words: ant colony algorithm, QoS routing, differential pheromone intensity, path mutated, twice ant colony algorithm search

摘要： 针对基本蚁群算法在求解QoS路由问题中存在的容易陷入局部最优和收敛速度慢的缺陷，提出一种基于信息差异度的蚁群算法对该问题进行求解。该算法在节点选择中嵌入路径信息素的差异度调节函数和迭代算子，动态调整节点选择策略;根据各路径上信息素的“集中”程度判断解的早熟、停滞情况，并引入路径变异和二次蚁群操作;根据最大-最小蚁群算法原理对信息素进行限制。仿真实验表明，算法全局搜索能力较强，能够跳出局部极值区间，快速收敛到全局最优解，算法是可行、有效的。

关键词: 蚁群算法, QoS路由, 信息差异度, 路径变异, 二次蚁群搜索