PRESENTED BY: VIJETA CHALLA PARNITHA KOTHAPALLY HIMA BINDU SALVAJI

 INTRODUCTION WHAT IS FAULT TOLERANCE WHY FAULT TOLERANCE  TYPES OF CONCERN OF FAULT TOLERANCE  MODELS  MAJOR AREAS OF FAULT TOLERANCE  IMPLEMENTATION OBJECT ORIENTED APPROACH FOR IMPLEMENTING ALGORITHM BASED FAULT TOLERANCE  CONCLUSION  REFERENCES  QUESTIONS

WHAT IS FAULT TOLERANCE Monitored and tolerated in order to provide a good performance for the users throughout the network. WHY FAULT TOLERANCE Reliability Survivability Performance

o Our goal is to maximize the network reliability at its design life. o As dependence on mobile terminals increases, users will demand the same system functionality, in terms of reliable service. o Reliability techniques comprise of the following phases. Fault confinement Fault detection Fault latency

Fault detection techniques Fault detection techniques Offline Detection Online Detection online detection techniques include parity checking, duplication and triplication. Stages: Diagnosis Reconfiguration Recovery Two most widely used recovery techniques are fault masking and retry.

Survivability is used to describe the available performance of a network after a failure. Strategies to improve network survivability can be classified into three categories: -> Prevention, -> Network design and capacity allocation, ->Traffic management and restoration. The ideal survivability goal is to make a network failure imperceptible to the user by providing service continuity and minimizing network congestion.

Survivability issues in wireless networks must take into account these unique characteristics: user mobility, power conservation in mobile terminals, security (encryption and authentication), the poor quality of radio links (in comparison to wired equivalents), channel capacity that is limited by a regulated frequency spectrum.

Survivability framework consists of three layers with survivability strategies possible at each layer. The three layers are termed  Access Layer Radio sub layer Link Layer  Transport Layer  Intelligent Layer Each of the three layers are characterized by network functions, network components, and communication links

Various issues and factors are considered for good performance of the network. In order to achieve over all good performance of the network fault tolerant techniques are implemented at various levels. For example, Recovery Recovery is a process where the type of the faults is identified and new clusters are allocated to the sensors. When the gateway is identified as failure all sensors in its clusters are recovered. The status message is parsed to extract the identity of sensors that cannot communicate with the gateway due to range faults in the gateways. Clustering is based on the distance between the sensors and gateways. During clustering each gateway creates a range set based on the communication range of the sensors and the gateways.

Fault models Fault detection and diagnosis Resiliency mechanisms MODELS: STUCK AT Value stuck at zero or one BRIDGING Physically connected with wired AND / wired OR depending on logic family.

PHYSICAL LAYER HARWARE COMPONENTS SYSTEM SOFTWARE MIDDLEWARE APPLICATION

OBJECT ORIENTED APPROACH FOR IMPLEMENTING ALGORITHM BASED FAULT TOLERANCE. The techniques which protect the integrity data items during storage, transmission and operations are referred to as algorithm based fault tolerance Algorithm-based fault tolerance techniques provide additional fault- tolerant layers that catch and handle faults which could not be corrected by the operating system and hardware. Object-based fault tolerance allows programmers to implement fault tolerance in their applications without having to master all the details of the discipline.

The fault tolerance is implemented as a firewall between the actual data object instance and the application, therefore isolating, detecting and correcting data errors before they propagate to other data objects. Layering of fault tolerance around data structures by implementing Weighted Checksum Codes (WCC) which encapsulates the standard matrix library Error-detecting and error-correcting codes, similar to Weighted Checksum codes, have been standard in data transmission and storage for many years which should be able to protect the information not only when it is being stored or transferred, but also when it is being modified. Weighted Checksum Codes is chosen because it provides an efficient implementation of fault-tolerant matrix objects.

Information hiding is a formal mechanism for restricting access to the internal data structures of the class and for protecting the internal state of the class object. An application creates matrix objects and manipulates the objects through the fault tolerance layer. The fault tolerance layer performs only error-detection and error-correction tasks and sends messages to the standard matrix object to perform the actual matrix operations. fault tolerance is transparent to the application.

Setting of an individual element of a matrix to a new value by the user. Every time an element is set to a new value, it must also update the corresponding checksums. Otherwise it would result in inconsistent checksums which would indicate the need to change the matrix item back to its previous value. Simply passing through the scalar-matrix operation message does not yield correct results.

Besides standard operators to access the matrix elements, methods are also implemented to update the checksums after each modification of a matrix element. The proper way of implementing scalar matrix addition or subtraction is to transform the scalar-matrix operation message into a matrix-matrix addition message. It is necessary to create a new check summed matrix of the scalar and send a matrix addition or subtraction message to the matrices

Study of fault tolerance in each layer and different type of implementations done to each layer and different algorithms proposed in order to over come faults in the network. One of the common implementation is object based approach for implementing algorithm based fault tolerance. Algorithm-based fault tolerance techniques can achieve high performance, tolerate physical failures in the system, and still produce correct results. object-based fault tolerance scheme that allows fault tolerance layers to be transparently added to data objects. C++ abstraction techniques are used to add algorithm-based fault tolerance to data objects without modification to the original object class or the application.

Fault tolerance in wireless sensor networks Farinaz Koushanfar1, Miodrag Potkonjak, Alberto Sangiovanni- Vincentelli1 Fault tolerance techniques for wirelss ad-hoc sensor networks Farinaz Koushanfar, Miodrag Potkonjak, Alberto Snagiovanni- Vincentelli Designing efficient fault tolerant systems in wireless networks, Guohong Cao Designing fault tolerant wireless access networks, C. Charnsprinyo, D. Tippe Fault Tolerant Deployment and Topology Control in Wireless Networks doi.ieeecomputersociety.org/ /HICSS

Additional factors that include for fault tolerant wireless network is a) Software b) Hardware c) Survivability d) Performance e) Heterogeneity of the network

At the application level, we expect that fault resiliency mechanisms required for common applications such as sensor fusion, data filtering and data aggregation will be of primary importance. It is important to observe that for each of these applications there will be a significant variety of approaches that will be used.

Q.What are the different types of algorithms? Distributed clustering algorithm Topology controlling algorithm Algorithm for improving the dependability

Duplication methods have 2 modules running on two different systems in a network then these two does the same computation and the results are produced and are compared, if the two messages do not agree on result, an error message is generated and mostly this method is usually used to detect a fault.

Traditionally, fault detection techniques are classified into offline and online detections. Most often for offline detection, special diagnostic program are employed either during idle periods of time or using multiplexing with a regular mode of operation. In an offline detection technique data is collected over a period of time and simulation program is run during the system idle time and the fault is detected.