In this paper, we proposed a hybrid self-healing mechanism with priority restoration for ATM networks, which integrates a backup-VP restoration method and a distributed flooding algorithm in the ATM layer, in order to meet different requirements of service classes Df survivability under minimum cost. This mechanism can restore higher priority VP's quickly by using preassigned backup VP's in case of single failure. It can also improve the restoration time of higher priority VPs in case of multi-failure. The flooding algorithm is adopted to restore the remain lower priority VP's father. This mechanism uses network resource more efficiently and obtains the better costperformance feature. tntroductionWith striding forward towards information society, the desires to increase the information transfer speed and capability inevitably leads to the development of new network: broadband ISDN. ATM is the target transfer mode for B-ISDN [lo], ATM switches are high-speed and highcapacity digital switches, if an ATM network element fails, its results are more serious than in existing network. Another way, digital fiber-optic technology has large capacity, plus advantages in lost, payload flexibility, and transmission quality, that make SDH the 'selected transport network for B-ISDN. Unfortunately optical cables do not have the resilience of copper cables in resisting cuts, so large capacity fiber is subject to frequent damage and results in more serious harm in the network. Moreover, with the advent of FITL, traffic concentrates into a finite set of more complex network element for economic reasons, the effects of this are to make individual element failure more catastrophic. These recent technological advances have made B-ISDN more vulnerable to single failures. In addition, telecommunication services are now an integral part of businesses, national security, and public health and safety. So the potential impact of single failure in 8-ISDN has never been greater on businesses, public perception, and geographic scope. Therefore, survivability of B-ISDN has become more and more urgent and important, ATM network should have restoration capability. If inevitable fibber cuts or network element ifails, the traffic affected by failure can be restored by switching to alternative VP route. So the effects of failure 01-1 ATM network traffic can be minimized in terms of traffic losses and time to recover. 0-7803-2916-3Fax: (86-10) 2283412A variety of distributed restoration algorithms for ATM network have been presented in Literature [I-5, 111. We try to integrate a backu -VP method and a distributed flooding algorithm in the ATM layer to obtain the better performance, and design a hybrid self-healing mechanism. Because both methods have thew advantages and disadvantages respectively, the backup VP method has advantages in restoration speed, standards and simplicity. The flooding algorithm has advantages in resource utilization. Before we describe the specification of the hybrid self-healing mechanism, we present service classes of survi...