points. The diagnosis proposed in this paper  is based on global sensitivity analysis method. Alsofuzzy logic is used to obtain the sensitivity curves. In an efficient method is applied in the selection of test nodes. This is done by searching for the minimumentropy index based on the available test points. Anefficient graph based method is proposed in .Thismethod can be used to select optimum test pointselection and also can be sued to build DFT. EfficientInclusion methods and Exclusion methods are proposed in  to select or de select test nodes, inother words the faster selection of optimum test points.A novel multi frequency approach is proposed in which drastically reduce the number of test frequenciesneeded to achieve maximum fault diagnosis. Thereduction achieved is better than any known methods.The method proposed in  consists of two parts.One is the creation of fault dictionary consisting of nominal and faulty states of the components andsecond is a novel fault detection and localizationalgorithm.This paper proposes a novel approach where both testnode and multi frequency techniques are used. Thisapproach is used to diagnose all the faults or thedesired number of faults.II Node-Frequency ApproachIn the analog fault diagnosis the prominent methodsused are multi node or multi frequency measurements.The research so far has been on developing methods tofind out optimum number of test nodes or testfrequencies that can identify the desired faults. This insome cases leads to more number of measurements being made thus drastically increasing the size of thedictionary.In this paper we have taken basically nodal analysisand then a choice of test frequencies is made based on. The proposed algorithm selects the nodes andfrequencies which isolate all or desired faults.In this paper two algorithms are presented. The firstalgorithm is for fault isolation and localization. Thesecond algorithm converts the integer coded faultdictionary into a binary dictionary which helps infaster fault isolation.The actual measurements of the CUT are noted downand these values are normalized if necessary. Fromthese values we form ambiguity sets. Now weconstruct another table called integer coded table usingambiguity sets. Then the original readings are replaced by integer numbers indicative of the ambiguity set towhich these values belong.The test frequency set is represented by f
, where N is the number of frequencies chosen.The nodes are represented by n
, where Prepresents the total number of nodes.The faults are represented by F
(nominal value) to F
,where N represents the total number of faults.
Select the test frequency set (f
Select the test nodes (n
) which areaccessible for each frequency.
Note the actual readings of the circuit for thetest frequency set and nodes chosen in steps 1 and 2.
Form the integer coded dictionary using theambiguity sets.
Identify unique integer codes called singletonsfor each row i.e. for each of the nodes selected.
Identify the node (n
) which has maximumnumber of singletons for a frequency f
, where 1<K≤Pand 1<J≤M. Select this node-frequency (n
) pair. If more than one node satisfies this condition, then go tostep 9.
If the number of singletons is equal to N+1,then go to step 12. If else go to next step 8.
Call Algorithm 2, to form binary dictionarywhich helps in identifying other nodes from theremaining (P-1) nodes belonging to the frequency f
,which can identify different faults. If all faults areisolated then go to step 12.
Find the total number of singletons for eachtest frequency. Then choose the node belonging to thefrequency which has the maximum number of singletons. If more than one frequency satisfies thiscondition choose any one of the nodes randomly.
If all the faults or desired number of faults arenot isolated, then repeat steps from 6 with the nexthighest number of singletons.
(IJCSIS) International Journal of Computer Science and Information Security,Vol. 8, No. 4, July 2010292http://sites.google.com/site/ijcsis/ISSN 1947-5500