Showing papers by "Nitte Meenakshi Institute of Technology published in 2014"

Effect of molybdenum doping on the electrochromic properties of tungsten oxide thin films by RF magnetron sputtering

Abstract: Thin films of pure and molybdenum (Mo)-doped tungsten trioxide (WO3) were deposited on indium tin oxide (ITO)-coated glass and Corning glass substrates by RF magnetron sputtering technique. The effect of Mo doping on the structural, morphological, optical and electrochromic properties of WO3 films was studied systematically. The energy dispersive X-ray analysis (EDAX) revealed that the films consist of molybdenum concentrations from 0 to 2 at.%. X-ray diffraction (XRD) studies indicated that with the increase of Mo concentration the structural phase transformation takes place from polycrystalline to amorphous phase. The crystallite size of the films decreased from 24 to 12 nm with increase of doping concentration of Mo in WO3. Scanning electron microscope (SEM) analysis revealed that Mo dopant led to significant changes in the surface morphology of the films. The electrochemical and electrochromic performance of the pure and Mo-doped WO3 were studied. The WO3 films formed with 1.3 at.% Mo dopant concentration exhibited high optical modulation of 44.3 % and coloration efficiency of 42.5 cm2/C.

Threshold Based Auto Scaling of Virtual Machines in Cloud Environment

Abstract: Cost effectiveness is one of the reasons behind the popularity of Cloud. By effective resource utilization cost can be further reduced and resource wastage can be minimized. The application requirement may vary over time depending on many factors (for instance load on the application); user may run different types of application (a simple MS word to complex HPC application) in a VM. In such cases if the VM instance capacity is fixed there is a high possibility of mismatch between the VM capacity and application resource requirement. If the VM capacity is more than the application resource requirement then resource will be wasted; if the VM capacity is less than the application resource requirement then the application performance will degrade. To address these issues we are proposing threshold based auto scaling of virtual machines in which VMs will be dynamically scaled based on the application resource utilization (CPU and Memory). Using our approach effective resource utilization can be achieved.

An intelligent system to detect, avoid and maintain potholes: A graph theoretic approach

Abstract: In this paper, we propose a centralized system which detects potholes on roads and assists the driver to avoid them. The system consists of 3 components namely sensors for detection, an inter-vehicle communication protocol to warn other vehicles about the new potholes which are detected, a central server which maintains a database of all the potholes. The system also consists of a GPS navigation system which will route the vehicle through an efficient path based on certain paramaters such as length of the route, intensity of potholes in the route, vehicle load and type, size limits of the road etc. In addition to this, the central server is configured to assign priority to the potholes and report high priority potholes to any local council such as a municipality so that they are repaired immediately.

FT2R2Cloud: Fault tolerance using time-out and retransmission of requests for cloud applications

Abstract: Ever increasing demand of distributed computing and Internet based services have become part and parcel of human daily life to perform activities Service failures occur due to failure of processing elements, storage, communication system, undetected/unexpected bugs in the service In interactive applications if a service takes more than the stipulated time or fail to meet hard and soft constraints as part of its response time then such scenarios are considered as service failure In this paper we propose algorithm to rank software components based on their reliability taking into account number of times a service time out and number of times it is being invoked by other software components In this work we measure the reliability of the software components in terms of number of responses as their throughput

Implementation of the MEMS-based dual-axis sun sensor for nano satellites

Abstract: This paper describes the complete design, characterization, and satellite integration of a miniaturized two axis sun sensor which will be used in the attitude control system of Project STUDSAT-2 (STUDent SATellite-2). The sun sensor based on Micro-Electro-Mechanical systems (MEMS) technology has a sun field-of-view greater than ±62°, with angle accuracy better than 0.5°. The characteristics of the micro sun sensor make it very effective in terms of its operation on a nano satellite. The complete design details along with the simulation and testing results of the sensor processing electronics have been included. The general requirements in designing the circuit were to minimize the power consumption and mass of the circuit. The components were chosen based on these criteria. The electronics involves current to voltage conversions along with unique multiplexing methods and normalization technique for any unwanted errors. The working of the on-board sun vector algorithm with its Matlab analysis results is mentioned.

Open source RTOS implementation for on-board computer (OBC) in STUDSAT-2

Abstract: STUDSAT-1 is the first pico-satellite developed in India by the undergraduate students from seven engineering colleges across South India. After the successful launch of STUDSAT-1 on 12th July, 2010, the team is developing STUDSAT-2 which is India's first twin satellite mission. STUDSAT-2 is undertaken by the undergraduate students from seven engineering colleges from the State of Karnataka, India. STUDSAT-2 consists of two Nano Satellites each having the dimension of 30 × 30 × 20 cm cube and weighing less than 10 kg. Each satellite has two payloads, one is the CMOS Camera with a resolution of 80 m and other is the ISL facility. The satellites are in along-the-track constellation architecture with the Master Satellite sending position data of the first image location on orbit and velocity of master satellite from GPS module to the Slave satellite via ISL for improving temporal resolution. On-board computer is the brain of a satellite whose operating system is designed to operate in resource constrained environments and is often tailored to the specific needs of its host system, while a real time operating system ensures that interrupts and other time critical tasks are processed when required. In order to maximize the accessibility the operating system must be inexpensive, and readily available, which demands an open-source OS. Implementation of a Real-Time Operating System (RTOS) provides several priority levels for tasks execution. FreeRTOS fulfills the requirements by having real-time capabilities, full availability of the source code, functional development environment, cross-platform support and community support. After an extensive literature survey, the ARM microcontroller with low power and high performance capability, STM32 ARM Cortex-M4 with 168 MHz CPU having 210 DMIPS is chosen as the main controller that supports FreeRTOS that can run on ARM Cortex-M cores. The On-Board Computer controls and coordinates the tasks of all other subsystems of the satellite. The different tasks executed by these subsystems have multi-threading and preemptive multitasking characteristics. A real-time operating system, FreeRTOS, uses advanced task scheduling techniques and a preemptive kernel, which allows multi-threading of processes to occur. This paper presents the technical justication for using FreeRTOS in STUDSAT-2. It presents implementation of FreeRTOS on STM32Cortex M4 controller which will increase the functional integration and performance of small satellite that simplifies software development, enables code modularity, and results in maintainable and expandable high-performance that will reduce the effort required to develop software for STUDSAT-2A/2B Mission.

A novel data mining approach for soil Classification

Abstract: Decision tree is a well known approach for classification in data mining. C4.5 and Classification and Regression Trees (CART) are two widely used decision tree algorithms for classification. The main drawback of C4.5 algorithm is that, it is biased towards attributes with more values while CART algorithm produces misclassification errors when the domain of the target attribute is very large. In view of these limitations, this paper presents a modified decision tree algorithm. The C4.5, CART and the proposed classifier are trained using data set containing soil samples by considering optimal soil parameters namely pH (power of Hydrogen), Ec (Electrical Conductivity) and ESP (Exchangeable Sodium Percentage). The model is tested with test data set of soil samples. The test proves that the modified decision tree algorithm has higher classification accuracy when compared to C4.5 and CART algorithms.

Trusted Neighbors Based Secured Routing Scheme in Wireless Sensor Networks Using Agents

Abstract: In Wireless Sensor Network (WSN) applications, nodes are deployed in public domains that are prone to be attacked by various types of attacks wherein the data confidentiality, integrity and authentication are compromised. In this paper, we propose Agent based SEcured Routing using Trusted neighbors (ASERT) in WSN. ASERT selects trustworthy neighbors and establishes secured routes using software agents. ASERT operates in following phases. (1) Defining Safeguard Agency (SA) and Routing Agency (RA), both consisting of static and mobile agents and a knowledge base. (2) Identification of trustworthy neighbor nodes by SA with trust model that consists of probability and Message Authentication Codes (MAC) model. (3) Probability model estimates the trustworthy channel and trustworthy nodes whereas MAC model ensures them. (4) MAC's are dynamically computed by agents by generating the keys with the help of Random Oracle Extractor (ROE). (5) RA establishes secured routes from source to the sink node using agents. (6) Agents effectively identify possible security threats on wireless channel and node. (7) An improvement in packet delivery ratio, packet delivery latency, route maintenance overhead, memory overhead and energy consumption is achieved in ASERT as compared to two protocols, namely, Bio-inspired Trust and Reputation model in WSN (BTRM-WSN) using ant colony based approach and Ambient Trust Sensor Routing that uses trust combined with location information for routing.

Piezoelectric acoustic pressure sensor diaphragm design for energy harvesting

Abstract: In this paper we present the effects of diaphragm design parameters on the efficiency of acoustic pressure sensor for energy harvesting. We have considered different design parameters of the diaphragm such as shape, dimensions, thickness, type of the piezoelectric material of the diaphragm. We studied the effect of changes in these parameters on the deflection, stress, strain, voltage generated and resonant frequency of the diaphragm which affect the efficiency of energy harvesting. The effect of the changes in these parameters on energy harvesting has been simulated and analyzed using Comsol Multiphysics tool.

Analysis and design of hysteresis current controlled multilevel inverter fed PMSM drive

Abstract: This paper deals with the modeling and analysis of multilevel inverter fed PMSM (Permanent Magnet Synchronous Motor) drive. The concept of hysteresis is applies to produce the controlled switching pulses to multilevel inverter and the concept of vector control is applied in PMSM to obtain linear dynamics similar to DC motor. The five level cascaded inverter fed PMSM drives are developed in MATLAB/SIMULINK environment using simpower system toolboxes. The performance of the proposed drive system is obtained under various operating conditions.

A Generic Agent Based Cloud Computing Architecture for E-Learning

Abstract: E-learning can be referred to as “education through electronic media”. It is one of the most emerging concepts in the field of technology. E-learning systems can be either synchronous or asynchronous. Agent based e-learning is helpful in managing the information overload, it can serve as academic expert and manages creation of programming environment for learners. There are many characteristics that an E-learning environment has to support; they are Interaction, Data Security, User Personalization, Adaptability, Intelligence, Interoperability, Accessibility and User Authentication. E-learning must also support a few other features like cost effectiveness, reusability, storage capacity, powerful computing and virtualization which can be provided by Cloud computing. Cloud computing is everywhere these days, pick any blogs, journals, papers. The Cloud computing Architecture is built using three models, Software as a Service (SaaS), Platform as a Service (PaaS) and Infrastructure as a Service (IaaS), which provides a variety of services. This proposed architecture integrates the features of Agents and Cloud Computing to produce an Agent based Cloud Computing Architecture to enhance e-learning.

Correlation of Tribological Properties with Microstructure and Mechanical Properties of Graphite Cast Irons Centrifugally Cast for Engine Liner

Abstract: This is a study of the influence of centrifugal process on the graphite morphology, mechanical and wear properties on flake graphite iron (FGI), spheroidal graphite iron (SGI) and compacted graphite iron (CGI). Melts of hypereutectic and almost of identical composition with or without melt treatment were centrifugally cast. The microstructure, mechanical and wear properties of these specimens were studied. In the microstructure for FG iron it has been observed that the centrifugal process produces flake size range class 2–3 (range 160–320 μm) and graphite of flake type A by about 67.9 % (field %) and combined flakes of type B, C, D and E will be of 32.1 % (field %). While SGI has been observed to have 96.1 % nodules and 330.0 nodules per square millimeter. Similarly CGI has been seen to produce 52.0 % nodules and 113 nodules per square millimeter. SGI possess the highest tensile strength, rupture strain and hardness of 604 N/mm2, 6.1 %, 233 BHN respectively. Whereas FGI possess the least tensile strength, rupture strain and hardness of 303 N/mm2, 0.65 %, 185 BHN respectively among the irons. CGI has a tensile strength, rupture strain and hardness of 369 N/mm2, 1.2 % and 200 BHN respectively which lies in between those of FGI and SGI. During the wear test similar materials for both disk and pin combination show higher co-efficient of friction and wear rate than those for dissimilar material combinations. SGI disk and FGI pin combination show least wear. This combination would be ideally suited for engine liner and piston rings.

Digital image processing technique for microstructure analysis of spheroidal graphite, iron

Abstract: In the present paper the microstructure analysis is done using the digital image processing technique. Digital image processing is a tool that permits fast, complete, and accurate acquisition of information. The processing of microstructure images is to sharpen the microstructure before quantitative analysis. Previously the analysis of the microstructure taken in analog form was performed manually. This procedure is time consuming for numerous set of images. To avoid this problem, Digital Image Processing techniques are being used. In this paper the main emphasis is given on the analysis of microstructure grain boundary, which is detected using various edge detection operators. Here the digital image is converted into a binary image using a threshold value to segment the image. By this procedure we can change the intensity of the image to study the properties of grains more accurately. This also helps for counting the spheroids in the microstructure. Also the image is enhanced by applying filtering technique. The results obtained from this approach are in the form of new microstructure image with smoothed grain areas and precisely detected grain boundary. Such effect allows optimizing further analysis of material structure to perform statistical calculations of average grain size or prepare material model for Finite Element Method simulations. The results of the analysis of microstructure images help us to correlate certain mechanical properties like ductility, malleability, brittleness etc.

Single event transient tolerant high speed phase frequency detector for PLL based frequency synthesizer

Abstract: In this paper Radiation Hardened by Design (RHBD) Technique is used to design a high speed Phase Frequency Detector (PFD). The PFD output is not related to input duty cycle. The PFD is tolerant to Single Event Transient (SET) effect. The design is implemented using TSMC 0.35μm CMOS process. It is used in harsh radiation environments suitable for high speed space applications.

Design and development of network-based consumer applications on Android

Abstract: This paper presents an innovative research approach to design and develop mobile consumer applications across communication network. The design involves mobile client that receives services provided by the base station server connected to database server. Target consumer applications implemented are Electronic Health Record (EHR) and Utility billing application. These applications are tested for correctness and analyzed for performance with varying input data type and size. The applications are also verified and validated by porting on a mobile device that runs Android operating system. The proposed research finds use in upcoming mobile based consumer applications for achieving increased cost-effectiveness, scalability and ease of use.

Bayesian Network Based Fault Tolerance in Distributed Sensor Networks

Abstract: A Distributed Sensor Network (DSN) consists of a set of sensors that are interconnected by a communication network. DSN is capable of acquiring and processing signals, communicating, and performing simple computational tasks. Such sensors can detect and collect data concerning any sign of node failure, earthquakes, floods and even a terrorist attack. Energy efficiency and fault-tolerance network control are the most important issues in the development of DSNs. In this work, two methods of fault tolerance are proposed: fault detection and recovery to achieve fault tolerance using Bayesian Networks (BNs). Bayesian Network is used to aid reasoning and decision making under uncertainty. The main objective of this work is to provide fault tolerance mechanism which is energy efficient and responsive to network using BNs. It is also used to detect energy depletion of node, link failure between nodes, and packet error in DSN. The proposed model is used to detect faults at node, sink and network level faults (link failure and packet error). The proposed fault recovery model is used to achieve fault tolerance by adjusting the network of the randomly deployed sensor nodes based on of its probabilities. Finally, the performance parameters for the proposed scheme are evaluated. Keywords—Bayesian network, distributed sensor networks, fault detection, fault tolerance, fault recovery, network control, routing.

Simulation and analysis of through silicon via (TSV) based inductance structures

Abstract: The multichip packaging techniques have evolved over the years and have resulted in many solutions towards 3D integrated circuits The key component to these technologies is Through Silicon Via Typical TSV structures (25 micron diameter and 20 micron length and 5 micron diameter and 50 micron length) are analyzed using HFSS, with a view to generate data and understanding of TSV performance, specific to realization of integrated passives TSV based inductor structure has been studied for different dimensions, number of turns and ground shielding This has been compared with those of 2D spiral structures also analyzed using HFSS Further, Via based vertical inductor realized within the metal layers of a sub micron technology has been simulated and analyzed for range of values and parasitics The parasitics of TSV based inductance are found to be comparable with those of conventional spiral structure; this is specific to resistance of L which is of the same order in both cases, for a given inductance value However, the area occupied on the die by a vertical TSV based inductor is considerably less than that by a conventional spiral inductor The range of values that can be achieved by a TSV based L depends on the TSV length, which in turn depends on the interposer thickness in chip / wafer stacking A 20 micron TSV based inductor has L in the range of 50–500 pH; a 50 micron interposer technology can generate much higher values of inductors With considerable advantage on the die size, the TSV based inductor is a potential integrated passive for stacked 3D microcircuits

Competitive performance analysis of two evolutionary algorithms for routing optimization in graded network

Abstract: In this paper we compare the two intelligent route generation system and its performance capability in graded networks using Artificial Bee Colony (ABC) algorithm and Genetic Algorithm (GA). Both ABC and GA have found its importance in optimization technique for determining optimal path while routing operations in the network. The paper shows how ABC approach has been utilized for determining the optimal path based on bandwidth availability of the links and determines better quality paths over GA. Here the nodes participating in the routing are evaluated for their QoS metric. The nodes which satisfy the minimum threshold value of the metric are chosen and enabled to participate in routing. A quadrant is synthesized on the source as the centre and depending on which quadrant the destination node belongs to, a search for optimal path is performed. The simulation results show that ABC speeds up local minimum search convergence by around 60% as compared to GA with respect to traffic intensity, and opens the possibility for cognitive routing in future intelligent networks.

CMOS architecture for SSTDR implementation

Abstract: Conventional Reflectomet methods were used for detecting faults on bundles of wires. These bundles of wires may be part of any super system like aircrafts, space shuttle, automobiles or nuclear plants. Faults include open and short circuits, partially open and partially short, frays, solder joints, and other small anomalies. Among the various Reflectometry methods, Spread Spectrum Time Domain Reflectometry (SSTDR) is practically feasible for a CMOS solution. SSTDR is found to be non-interfering with digital signals. Here, we are proposing a CMOS architecture which implements the SSTDR algorithm. The proposed CMOS architecture consists of both analog and digital modules which can be developed into an integrated circuit which can automatically detect faults on wires with precisions on the order of a few inches.

Ground based modeling and real time On-board calibration of a three axis magnetometer

Abstract: This paper presents a dynamic and robust method to model and calibrate a three axis magnetometer. The STUDSAT-2 (STUDent SATellite-2) satellite uses a low cost three axis digital magneto resistive magnetometer. Low cost magnetometers can be used for estimating the orientation with respect to the magnetic North. Although magnetometers work very well in clean environment, they are strongly influenced by magnetic perturbations produced by the electronics around them. Modelling the biases and scale factors of the magnetic sensor is a viable solution to estimate the errors. The local geomagnetic field vector is measured with a magnetometer and by comparing it to a model of the earth's magnetic field, an estimation of the attitude angles and angular rates can be made. A control torque to change the attitude of a satellite can be generated by a magnetic moment that interacts with the local geomagnetic field. On board calibration of the sensor helps in increasing the accuracy of spacecraft's attitude.

Abnormality Detection in Ovarian Ultrasound Images using Active Contours

Abstract: Analyzing ovarian size parameters and its shape is the first hand information required in Ovary diagnosis. Automated segmentation of Ovary and measuring its size and analyzing its shape can help doctors in mass screening programme in faster diagnosis. In this paper, we propose a new improved algorithm (anisotropic diffusion filter, CLAHE enhancement, and active contour segmentation) to extract the Ovary from its homogeneous background followed by geometrical and shape feature extraction of the Ovary for automated computer-assisted analysis. The proposed algorithm is tested on 75 Transvaginal ultrasound images of ovaries. The experimental results are validated against the manual measurements done by the expert and the results obtained by our algorithm are in good agreement with experts inputs. The proposed algorithm could achieve an average Error Percentage EM1 of 4% for Major-Axis length and average Error Percentage EM2 of 5.43% for Minor-Axis length and classification rate of 91.66%.

Effect of Online Reviews on Customers of Bangalore-Based Real Estate Company - An Exploratory Study

Abstract: Social media, consumer forums, review sites and blogs facilitate wider and faster sharing of customer experiences. Consumers look at user-generated reviews in consumer portals before deciding to purchase products/services. This Paper aims at studying effects of online reviews on intentions of existing and potential customers to pursue specific actions with regard to budget- and mid-range housing projects offered by a Bangalore-based realty company (Bangalore Realty Limited - BRL). Main data set for the study comprises of 221 user online reviews - in eight consumer review/complaint portals over a period of four years (2010-2014), supplemented by personal interviews of 25 customers. Online reviews are subjected to manual content analysis. Emergent themes in the reviews are identified and frequencies for negative and positive themes, valence and outcomes - which culminate in actions such as ‘taking legal recourse’, ‘withdrawal from the project’ or intentions ‘not to recommend’ and ‘to recommend’ - are determined. Such intentions expressed in reviews convey strong messages as further “threads” and “comments” are generated. Parameters like ‘Mean time to get the complaints resolved’ (in case of registration of sites and getting refunds on cancellation) are determined. This study reveals that user-generated online reviews help customers for making decisions about ‘continuance’ or ‘withdrawal’ or ‘staying away’ from the projects of BRL. At the same time BRL itself benefits from monitoring and responding to such reviews through setting up of appropriate customer complaints handling system, for gathering, transmitting and resolution of issues raised in online reviews, which would minimize harmful effect of negative reviews and maximize beneficial effect of positive reviews. Giving timely responses to customer complaints/online reviews, and periodic communication about project status may not require additional capital outlay or human resources. BRL’s current policy of ignoring or being indifferent to online reviews would be detrimental for its revenues, reputation and future growth. Study is based on analysis of qualitative data which cannot be subjected to robust statistical tests. Many critical dimensions such as customer/user preferences, motives, expertise, involvement and credibility (of the portals, users posting online reviews) are not analyzed in this study, as online reviews do not contain relevant data.

Power management of hybrid fuel cell and microturbine based DG system in utility connected mode

Abstract: The distributed generation (DG) with fuel cell and microturbine generation (MTG) system is an emerging approach to provide a quality power in the distribution network though the renewable and nonrenewable energy resources. Thus, interconnection of fuel cell MTG based DG system to the distribution network has to withstand several grid issues. This paper presents the hybrid operation of fuel cell and MTG based DG system in utility interconnected mode. The hybrid DG system includes a solid oxide fuel cell (SOFC) and MTG system with DC link based air and fuel flow control strategy to limit their output power. The hybrid SOFC and MTG system are integrated at common at common DC link and is delivered to grid through the power electronics interfacing. The model of hybrid DG system with SOFC and MTG system is presented using Matlab/Simulink environment and it performance has been studied through the simulation.

Optimal path selection in Graded network using Artificial Bee Colony algorithm with Agent enabled Information

Abstract: In this paper we propose a network aware approach for routing in graded network using Artificial Bee Colony (ABC) algorithm. ABC has been used as a good search process for optimality exploitation and exploration. The paper shows how ABC approach has been utilized for determining the optimal path based on bandwidth availability of the link and how it outperformed non graded network while deriving the optimal path. The selection of the nodes is based on the direction of the destination node also. This would help in narrowing down the number of nodes participating in routing. Here an agent system governs the collection of QoS parameters of the nodes. Also a quadrant is synthesized with centre as the source node. Based on the information of which quadrant the destination belongs, a search is performed. Among the many searches observed by the onlooker bees the best path is selected based on which onlooker bee comes back to source with information of the optimal path. The simulation result shows that the path convergence in graded network with ABC was 30% faster than non-graded ABC.

Software Based DFS Technique for Parallel Applications to Conserve the Power

Abstract: Scientific parallel applications need more computing power, to produce such computing power we need high end hardware. This hardware consumes more power and generates lot of heat. Cooling system is required to make sure that the hardware is healthy otherwise it will be prone to failure. As a result, cost of data center operation will be more and it will not be environmental friendly. By reducing the power consumption we can reduce the cost as well as make data center environmental friendly. This work proposes software based Dynamic Frequency Scaling technique which helps in reducing the power consumption by parallel applications. To evaluate our technique we used popular NAS parallel benchmark and observed that up to 40% energy saving is achieved.

A QoS-based routing approach using genetic algorithms for bandwidth maximisation in networks

Abstract: This paper addresses the path selection problem from a known source to the destination in dense networks. The proposed solution for route discovery uses the genetic algorithm approach for a QoS-based network. The multi point crossover and mutation helps in determining the optimal path and alternate path when required. The input to the genetic algorithm is a learnt module which is a part of the cognitive router that takes care of four QoS parameters. Here, the set of nodes selected for routing is determined by delay, jitter and loss. On this graded surface of nodes selected, the bandwidth parameter is considered for path selection. The aim of the approach is to occupy the maximised bandwidth along the forward channels and minimise the route length. The population size is considered as fixed nodes participating in the network scenario, which will be limited to a known size of topology. The simulated results show that by using genetic algorithm GA approach the probability of convergence to shortest path is higher.

An intelligent routing approach using genetic algorithms for quality graded network

Abstract: Intelligent routing in networks has opened up many challenges in modelling and methods, over the past decade. Many techniques do exist for routing on such an environment where path determination was carried out by advertisement, position and near-optimum node selection schemes. In this paper, an efficient routing scheme has been proposed using genetic algorithm for a grade-based two-level node selection method. This method assumes that nodes have the knowledge of its environment and is capable of taking decision for route discovery. The data learnt from the topology which is under consideration for routing, is saved in its local memory. In this two-level node selection scheme, the route discovery operation takes place in multiple levels. At the first level, the grade based selection is applied for considering the most optimal nodes which would be fit for sending data. At the second level, the optimal path is discovered using Genetic Algorithm. The simulation result shows that faster convergence of path took place in the case of the proposed method with good fitness value, as compared to non-graded network.