ITR/SI: On Robust and Secure Mobile Ad Hoc and Sensor Networks

移动自组织网络（MANET）是一种由多跳无线链路连接的移动自治节点系统，没有底层基础设施。由于其对人类生活的深远影响，其研究受到了前所未有的重视。MANET中最重要的问题是路由、服务质量、安全、电源管理和交通和移动性建模，我们计划解决其中的许多问题。在MANET中，当移动节点有数据包要发送时，它首先查询路由缓存以确定它是否已经有到目的地的路由，否则它将通过广播启动路由发现。线路一旦建立，它的延续和维护是有效通信的首要问题。路由可能因移动性而中断，并由路由错误包决定。我们采用局部路由修复（LLR）方法来解决这个问题，而不是采用传统的全局重新淹没方法来进行新的路由搜索，并以最小的开销实现重新连接。需要进一步的工作来扩展方案，增加修补部分的长度。当单向链路存在时，我们期望这在路由开销和延迟方面产生更高的节省。我们还计划为不同的协议定义分析框架，并确定它们的性能。在MANET中对调度的需求主要是由于资源的可用性有限而产生的。我们计划研究数据包级调度策略的影响，该策略基于从传输数据的特征派生的度量来传输数据包，如数据包的长度或数据包所遍历的路由的长度，甚至数据包的优先级。我们计划测试其他路由方案，并定义和制定每个节点数据包队列的分析模型，以便我们可以确定不同类型数据包的端到端延迟。manet的有效运行依赖于以分布式方式维护适当的路由信息。路由信息的保密性非常重要，因此不仅可以安全地交换有效载荷数据，而且可以安全地交换所携带的路由消息头和移动节点的位置信息。我们定义了一个可以防止外部攻击的身份验证方案。只有在收到路由应答时，我们才能在应答路径上使用质询-响应机制执行相互身份验证，从而最大限度地减少大量开销。我们计划提出一个分布式入侵检测系统来处理它们。我们将根据广泛研究交通模式后得到的交通概况来建立这些系统。我们计划进行详细的模拟，以获得有助于我们区分有效和恶意数据的指标。无线传感器网络可以很容易地部署，没有任何安装成本或预先计划的组织。为了提高容错性，可以使用数千个这样的传感器。为无线自组织网络定义的传统路由协议不能很好地扩展传感器网络，因为它们与传统网络不同，是以数据为中心和特定于应用程序的。大多数传感器网络协议定期从环境中收集数据。我们认为存在着一种网络的需求，这种网络能够对感知属性的变化做出即时反应。最近，我们为这样的网络提出了一个叫做TEEN的协议。我们还在寻找将主动协议和被动协议的最佳特性结合起来形成混合网络的方法。我们还认为，传感器网络应该为最终用户提供动态控制能源效率，准确性和响应时间之间的三方权衡的能力。简而言之，该提案研究将对manet和传感器网络的鲁棒性和安全性产生长期影响。

A Mobile Ad hoc NETwork (MANET) is a collection of mobile autonomous system of nodes connectedby wireless links with multi-hop communication and no underlying infrastructure. Research in the field ofMANET is receiving unprecedented attention due to their profound impact on human life. The mostimportant problems in MANET are Routing, Quality of Service, Security, Power management and Trafficand Mobility modeling and we plan to address many of these issues.In a MANET, when a mobile node has a packet to send, it first consults its route cache to determinewhether it already has a route to destination, else it initiates route discovery by broadcasting. Once a routeis established, its continuation and maintenance is of prime concern for efficient communication. A routecould be broken by mobility and is determined by route error packets. Rather than employing aconventional way of global re-flooding for a new route search, we address this by Local Route Repair(LLR) approach and achieve re-connectivity with minimum overhead. Further work is needed to extendour scheme by increasing the length of patch-up part. We expect this to yield a higher amount of savingsin terms of routing overhead and delay when unidirectional links are present. We also plan to define ananalytical framework for different protocols and determine their performance.The need for scheduling in a MANET arises primarily due to limited availability of resources. We plan tostudy the impact of packet level scheduling strategies that transmit packets based on a metric derivedfrom characteristics of the data being transmitted, like the length of the packets or the length of the routebeing traversed by the packets or even the priority of the packets. We plan to test other routing schemesand define and formulate an analytical model for packet queue at each node so that we could determineend-to-end delay of different types of packets.Effective operation of MANETs is dependent on maintaining appropriate routing information in adistributed fashion. Confidentiality of routing information is important so that not only the payload databut also the routing message headers carrying, the location information of the mobile nodes can beexchanged securely. We have defined an authentication scheme that can prevent the external attacks. Weminimize substantial overhead by performing mutual authentication using challenge-response mechanismalong the reply path only when the route reply is received. We plan to propose a distributed intrusiondetection system to handle them. We will build these systems based on traffic profiles that we wouldobtain after extensive study of the traffic patterns. We are planning to perform a detailed simulation toobtain metrics that would help us in differentiating valid and malicious data.Wireless sensor networks can be easily deployed without any installation costs or pre-plannedorganization. To increase fault tolerance, thousands of these sensors might be employed. The traditionalrouting protocols defined for wireless ad hoc network do not scale well for sensor networks as they aredata centric and application-specific unlike traditional networks. Most protocols for sensor networkcollect data periodically from environment. We feel that there exists a need for networks geared towardsresponding immediately to changes in the sensed attributes. Very recently, we have proposed a protocolcalled TEEN for such networks. We are also looking at ways to combine the best features of bothproactive and reactive protocols to form a hybrid network. We also believe that sensor networks shouldprovide the end user with the ability to control the three-way trade-off between energy efficiency,accuracy and response time dynamically. In brief, the proposal research will have a long-term impact onthe robustness and security of MANETs and sensor networks.