Top Banner
RTOS PORTING ON SMALL FOOT PRINT ARM PROCESSOR AND IMPLEMENTATION OF NETWORK PROTOCOL By SAMADHAN D. MALI Under the Guidance of PROF. Dr. AJAY D. JADHAV Department of E&TC Sinhgad College of Engineering
88

arm processor

Nov 03, 2014

Download

Documents

Shivani Paliwal

its a presentation on arm processor which can help in the priject
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: arm processor

RTOS PORTING ON SMALL FOOT PRINT ARM PROCESSOR AND IMPLEMENTATION OF

NETWORK PROTOCOL

BySAMADHAN D. MALI

Under the Guidance ofPROF. Dr. AJAY D. JADHAV

Department of E&TC

Sinhgad College of Engineering

Page 2: arm processor

ContentsReview of Theories ObjectiveIntroductionExperimentationResults & DiscussionsConclusionPublicationsReferences

Page 3: arm processor

Review of TheoriesRTOS

Real Time Operating SystemDeterministic NatureServices

Task ManagementIntertask Communication and SynchronizationI/O SupervisorDynamic Memory AllocationTimers

Page 4: arm processor

Review of TheoriesRTOS Scheduler

1. Non-preemptive scheduling

ISRISR makes high priority task ready

Low priority task relinquishes CPU

Low priority task

High priority task

Page 5: arm processor

2. Preemptive scheduling

ISR

High priority task relinquishes CPU

ISR makes highpriority task ready

Low priority task

High priority

task

Review of Theories

Page 6: arm processor

Review of TheoriesNETWORK PROTOCOLS

MODBUS PROTOCOL Present at Second level of ISO-OSI Model It’s a Serial Line Master-Slave Protocol Query/Response Mode of Communication

Page 7: arm processor

Review of Theories Message Frame Structure

Page 8: arm processor

ObjectiveDesigning embedded system has

constraints like memory space, energy consumption, reliability, performance, execution time of an application, small size and capability to upgrade software.

In the view of designing embedded system application where the scheduling of multiple tasks is required, the real time operating system is generally used as abstraction layer between hardware and application software. To achieve this, it aimed to design an ARM processor based system for implementation of MODBUS protocol integrating with RTOS µCOS-II.

Page 9: arm processor

Block Diagram of the Project

Introduction

Page 10: arm processor

Introduction

Page 11: arm processor

Basic structure of circular microstrip antenna is

Present Theories

Page 12: arm processor

Microstrip antennas have Narrow bandwidth Poor polarization purity

Problem Statement

User
EXPLAIN IN DETAIL. WHAT WOULD BE EFFECTS?
Page 13: arm processor

ObjectivesTo enhance the bandwidth of circular

microstrip antenna.

To design circularly polarized circular microstrip antenna

Page 14: arm processor

Methodology (Detailed)

1. Bandwidth Enhancement Techniques1. Modification of shape

2. Use of stacked structure

3. Use of parasitic patches

CMSA with Two Notches CMSA using Stacked Structure

Use of Planer Multiresonator

Patches

User
WHY DID I SELECTED FIRST ONE? WHY NOT OTHERS?Second increases height of structure, Not desireable in confirmal applications.Increase back radiation if aperature coupledThird- large size, variation in radiation patterns
Page 15: arm processor

PolarizationOrientation of the electric field (E-plane) of the radio

wave with respect to the earth's surface

Linear Polarization- Only one component presentCircular Polarization

When Ex0 = Ey0 and 900 phase difference between them.Measured in terms of Axial Ratio

Otherwise wave is elliptically polarized. 𝐴𝑅= 𝑚𝑎𝑗𝑜𝑟 𝑎𝑥𝑖𝑠𝑚𝑖𝑛𝑜𝑟 𝑎𝑥𝑖𝑠

Page 16: arm processor

Circular Polarization (CP)Advantages

No requirement of alignment between transmitter and receiver antenna

After reflection from metallic objects, sense of polarization reverses, so useful in RADAR, navigational systems

Techniques to produce CP

Dual Feed Single Feed

Page 17: arm processor

Conditions for CP1. E-field must have two orthogonal components

2. These components should have equal magnitude

3. Orthogonal components must be 900 out of phase

Page 18: arm processor

Methodology for CP (Selected) Modification of shape

CMSA with two notches

Page 19: arm processor

Design of Simple CMSAFor Simple CMSA

Effective radius (ae ) of the patch in CMSA is calculated as

Where Knm : mth zero of first order derivative of Bessel’s function ≈ 1.84118

c : Velocity of light

f0 : Resonant frequency

ɛe : Effective dielectric constant

User
WHAT IS SIGNIFICANCE OF BESSEL'S FUNCTION?
Page 20: arm processor

But actual patch radius is required, which can be calculated as

Where

a : actual radius

h : Height of dielectric substrate

ɛr : Dielectric constant

Design of Simple CMSA

User
whether to mention reference here
Page 21: arm processor

Specifications : Input

1. Dielectric constant= 4.4 (Material chosen- Glass Epoxy)

2. Height of substrate= 1.6 mm Output

1. Resonant Frequency = 2.45 GHz

2. Bandwidth >= 2%

3. Gain ≈ 2 dBi

Design of Simple CMSA

User
why
Page 22: arm processor

Steps1. For given input parameters, effective patch radius is

ae = 17.1 mm

2. From ae , actual patch radius is

a= 16.6 mm

Design of Simple CMSA

Page 23: arm processor

Design of CMSA for CP

Change in geometry by amount ∆𝑆

Page 24: arm processor

For minimum AR

Where K11 = 1.84118

∆𝑆𝑆 = 0.543𝑄0

Δ𝑆𝑆 ∗𝑄0 ≈ 1𝐾11

Design of CMSA for CP

Page 25: arm processor

Q0 can be calculated from

For our design

Q0 ≈ 46.15 but

So

ℎ𝜆0 =0.013

𝑄0ሺ𝜀𝑟ሻ= 𝑄0 ∗ξ𝜀𝑟ξ2.55

Q0 (𝜀𝑟) = 60.62

Design of CMSA for CP

Page 26: arm processor

So now

But

This is perturbation area.

∆𝑆𝑆 = 8.956e-3

𝑆= 𝜋∗𝑎2 𝑆= 8.657e-4

∆𝑆=7.753e-6≈ 8 mm2

Design of CMSA for CP

Page 27: arm processor

Design Parameters for Modified CMSA

Antenna Parameters

Circular patch radius

Coaxial feed position

Coaxial feed probe radius

Perturbation Area Length

Perturbation Area Width

Page 28: arm processor

OptimizationDetermining best course of action amongst different

alternatives available in decision making.

ORProcess of finding optimal value of objective function

under given set of constraintsPhases in optimization

1. Modelling

2. Solution of mathematical model

3. Validation of results

If results are not up to the mark

Page 29: arm processor

Need of OptimizationIn EM, equations are highly nonlinear and complexLarge number of optimization parametersOutput parameters depends upon complex relations

between input parameters.

Out of available combinations of input parameters, which would be best?

Page 30: arm processor

Optimization Techniques

Random Search TechniquesSimulated AnnealingGenetic AlgorithmFuzzy LogicArtificial Neural Network

Page 31: arm processor

Random Search Based TechniquesGenerates random sequence of input parameters and

returns best point

….where value of objective function is lowest

Not suitable if parameters are many And if no fix relation between input parameters

Page 32: arm processor

Simulated AnnealingIn each generation, current solution changed randomly.

X -> X’∆f=f(X)-f(X’)If ∆f<=0, Change is acceptedIf ∆f>=0, change is accepted with conditionsDisadvantages

Quality of solutions is not stable..Computational time is quite long.

Page 33: arm processor

Genetic Algorithm

Globally accepted parameter optimization techniqueBased on Darwin’s concept

“Survival of the fittest”GA’s are known as the best approach if the number of

unknown parameters increases.

Page 34: arm processor

Algorithm in Detail…….

Establish encoding/decoding of parameters

Generate M random chromosomes

Evaluate cost function for the chromosomes

Rank chromosomes

Discard inferior chromosomes

Mate remaining chromosomes

Done? Mutations

Stop

Start

No

Yes

• Very slow search speedDue to selection of inappropriate parameters• Premature convergenceParameters always remain fixed irrespective of operating environment

Page 35: arm processor

Fuzzy LogicGeneralization of Boolean logic implementing the

concept of partial truth or uncertainty.Fuzzy logic is conceptually easy to understand.Fuzzy logic can be built on top of the experience of

experts.Not so effective when physical processes relationships are

not fully understood.

Page 36: arm processor

Artificial Neural NetworkA computational model inspired from Human

Nervous System.Learns from examplesConsists of an interconnected group of artificial

neuronsProcesses information using a connectionist approach

to computation.Resembles human brain in following aspects

1. Knowledge acquisition through learning process

2. Interneuron connection strengths used to store knowledge

Page 37: arm processor

A Simple Artificial Neural Net

x1

x2

y2

Input Layer Output Layer

W1 (Weights)

W2x2

x1

Inpu

ts

(Output)

Page 38: arm processor

Why ANN ?Massive parallelismDistributed representation and computationAbility and adaptability to learnGeneralization ability Expertise in perceptual problemsEase of implementation

Page 39: arm processor

ANN BasicsTypes of ANN

Feed BackFeed forward CompetitiveRecurrent

Training Methods1. Supervised 2. Unsupervised3. Reinforcement

Page 40: arm processor

ANN BasicsTypes of ANN

Feed BackFeed forward CompetitiveRecurrent

Training Methods

1. Supervised 2. Unsupervised3. Reinforcement

Page 41: arm processor

Multilayer Perceptron…….Feed forward network with supervised learningGoal of this type of network is to create a model that

correctly maps the input to the output using historical data

Single Hidden Layer Multiplayer Perceptron Network

Input Parameters

Hidden Layer

Output Parameters

Page 42: arm processor

Learning of Multilayer Perceptron

Back Propagation Algorithm

Page 43: arm processor

Application of ANN to Project

Input and Output Parameters

Sr. No. Input Parameters Output parameters

1 Circular patch radius Bandwidth

2 Coaxial feed position Resonant Frequency

3 Coaxial feed probe radius Axial ratio

4 Perturbation Area Length Antenna Efficiency

5 Perturbation Area Width Gain

Page 44: arm processor

ANN Models (Synthesis)

Artificial

Neural

Network

Model

Patch Radius

Feed Location

Feed Probe Radius

Perturbation Area Length

Perturbation Area Width

Bandwidth

Resonant Frequency

Axial Ratio

Antenna efficiency

Gain

Page 45: arm processor

ANN Models (Analysis)

Artificial

Neural

Network

Model

Patch Radius

Feed Location

Feed Probe Radius

Perturbation Area Length

Perturbation Area Width

Bandwidth

Resonant Frequency

Axial Ratio

Antenna efficiency

Gain

Page 46: arm processor

Before Using ANNWe should decide1. Number of input and outputs in single training pair

2. Total number of training pairs

3. Number of hidden layers and neurons in each hidden layer

Page 47: arm processor

Design Considerations for ANN

Artificial

Neural

Network

Model

Patch Radius

Feed Location

Feed Probe Radius

Perturbation Area Length

Perturbation Area Width

Bandwidth

Resonant Frequency

Axial Ratio

Antenna efficiency

Gain

Synthesis ANN Model

1. Number of inputs and outputs in 1 training pair

Page 48: arm processor

Design Considerations for ANN

2. Total number of training pairs

Page 49: arm processor

Database for ANN

Input parameter and no. of combinations

Parameter No. of combinations

Patch radius 3

Parameter Initial value Final value Step

Patch radius 16.55 mm 16.65 mm 0.5 mm

Actual values considered

Page 50: arm processor

Database for ANN

Input parameter and no. of combinations

Parameter No. of combinations

Feed position 3

Parameter Initial value Final value Step

Patch radius 5 mm 6 mm 0.5 mm

Actual values considered

Page 51: arm processor

Database for ANNInput parameter and no. of combinations

Parameter No. of combinations

Feed probe radius 4

Parameter Initial value Final value Step

Patch radius 0.5 mm 0.8 mm 0.1 mm

Actual values considered

Page 52: arm processor

Database for ANNInput parameter and no. of combinations

Parameter No. of combinationsLength and width of Perturbation Area

4

Actual values considered

Sr. No.Length of P. A.

(in mm)Width of P. A.

(in mm)

1 1 4

2 2 2

3 3 1.33

4 4 1

Page 53: arm processor

Database for ANN

Total combinations

Total combinations- (3*3*4*4)=144

Parameter No. of combinations

Patch radius 3

Feed position 3

Feed probe radius 4

Length and width of P. A. 4

Total Number of Training Pairs=144

Page 54: arm processor

Database

Page 55: arm processor

Design Considerations for ANN

2. Total number of training pairs = 98 Validation Samples= 21 Test Samples= 21

3. Number of Hidden Layers and

Number of Neurons in Each Hidden Layer

Page 56: arm processor

Feed forward network

Input Parameters

Hidden Layer

Output Parameters

Design Considerations for ANN

Page 57: arm processor

Training ParametersNo. of epochs- 5000 (max)Goal = 1e-5

Maximum Number of weights (w) = 95

Design Considerations for ANN

3. Number of Neurons in Each Hidden Layer

Page 58: arm processor

Structure of Input Layer

5 weights for 1 neuron; 5 Neurons in I/P Layer;Total Weights = 25

Page 59: arm processor

Structure of Hidden Layer

5 neurons in previous layer; 7 Neurons in Hidden Layer;Total Weights = 35

Page 60: arm processor

Structure of Output Layer

7 neurons in previous layer; 5 Neurons in O/P Layer;Total Weights = 35

Page 61: arm processor

Structure of Total Network

Input Layer Hidden Layer Output Layer

Page 62: arm processor

Performance (Synthesis ANN Model)

Sr. No.

No. of Neuronsin Hidden

Layer

No of Epochs

Required

Average Percentage Deviation

Training Validation Testing

1 7 48 0.078 0.7 0.61

2 6 33 0.2042 0.3638 1.088

3 5 30 0.0595 0.1342 0.5667

4 4 87 0.0802 0.8796 0.7787

5 2 13 1.0872 4.3304 5.1172

Forward Training, Validation and Testing

Page 63: arm processor

Sr. No.

ParameterInput Vector

TargetVector

Output VectorGiven by ANN

Percentage Deviation

1 Patch Radius 16.55 89 88.4 0.7068

2 Feed position 5 2.43 2.43 0.0037

3 Feed Probe Radius 0.7 3.07 3.00 2.1737

4 Length of P.A. 3 36.37 36.47 0.2839

5 Width of P.A. 1.33 1.956 1.993 1.9086

Average Percentage Deviation=0.1384

Sample Number: 75

Results (Synthesis ANN Model)

Artificial

Neural

Network

Model

Patch Radius

Feed Location

Feed Probe Radius

Perturbation Area Length

Perturbation Area Width

Bandwidth

Resonant Frequency

Axial Ratio

Antenna efficiency

Gain

Page 64: arm processor

Performance (Analysis ANN Model)

Sr. No.

No. of Neuronsin Hidden

Layer

No of Epochs

Required

Average Percentage Deviation

Training Validation Testing

1 7 29 0.1763 0.4520 0.7160

2 6 24 0.5362 0.7756 1.0259

3 5 15 0.7711 1.1166 1.4653

4 4 38 0.6816 0.8573 1.2640

5 2 35 1.1512 4.0969 5.0671

Reverse Training, Validation and Testing

Page 65: arm processor

Sr. No.

ParameterInput Vector

TargetVector

Output VectorGiven by ANN

Percentage Deviation

1 Bandwidth 81 16.65 16.62 0.4225

2 Resonant Freq. 2.427 5.5 5.425 1.3712

3 Axial Ratio 3.08 0.8 0.8376 4.669

4 Efficiency 35.19 3 2.994 0.2125

5 Gain 1.81 1.33 1.331 0.0612

Results (Analysis ANN Model)

Average Percentage Deviation= 1.3533

Sample Number: 80

Artificial

Neural

Network

Model

Patch Radius

Feed Location

Feed Probe Radius

Perturbation Area Length

Perturbation Area Width

Bandwidth

Resonant Frequency

Axial Ratio

Antenna efficiency

Gain

Page 66: arm processor

Sr. No.

PerformanceParameter

Desired value

Physical parameter

Result given by ANN ( in mm )

1Bandwidth 110 MHz Patch Radius 16.48

2Resonant Frequency 2.45 GHz Feed position 5.93

3Axial Ratio 1.2 Feed Probe Radius 0.4075

4Antenna Efficiency 36 % Length of P.A. 4.09

5Gain 2 Width of P.A. 0.9032

Results (Analysis ANN Model)

Page 67: arm processor

Results Given by IE3D Model

Parameter Expected value

Simulated value Percentage Deviation

Bandwidth 110 MHz 112 MHz -1.81

Resonant frequency 2.45 GHz 2.449 GHZ 0.04

Axial ratio 1.2 1.26 -0.05

Antenna efficiency 36 % 36.66 % -1.833

Gain 2 2.02 1

Average Percentage Deviation= 0.9446

User
Why gain is decreased??Eficiency increased??
Page 68: arm processor

SimulationSoftware chosen

IE3D by Zeland® corporationBased on method of momentsHigh efficiency, high accuracy and low cost

electromagnetic simulation tool on PCs with Windows based graphic interface.

Built-in library for construction of complicated structures, such as circles, rings, spheres, rectangular.

3D and 2D display of current distribution, radiation patterns

Page 69: arm processor

Simulation of simple CMSA

Simple CMSA geometry in IE3D MGrid

Page 70: arm processor

Results of Simple CMSA

S11 vs Frequency

CMSA S11 (minimum)

Resonant Frequency

Simple -23 dB 2.449 GHz

Page 71: arm processor

VSWR vs Frequency

Results of Simple CMSA

CMSA VSWR

Simple 1.149

Page 72: arm processor

Smith Chart

Results of Simple CMSA

Page 73: arm processor

Resonant

Frequency

(in GHz)

Bandwidth( in MHz)

Bandwidth(in %)

AxialRatio

Ant.Efficiency

Gain

2.449 56 2.28 46.5 37.55 2.09

Results of Simple CMSA

Page 74: arm processor

Simulation Results Comparison(Simple CMSA & Modified CMSA)

CMSA VSWR

Simple 1.149

Modified 1.06

VSWR Plot

Page 75: arm processor

Simulation Results Comparison(Simple CMSA & Modified CMSA)

CMSA S11 (minimum)

Resonant Frequency

Simple -23 dB 2.449 GHz

Modified -30.71 dB 2.449 GHz

S11 Plot

Page 76: arm processor

Simulation Results Comparison(Simple CMSA & Modified CMSA)

Smith Chart Plot

Page 77: arm processor

Simple CMSA Modified CMSA

Simulation Results Comparison(Simple CMSA & Modified CMSA)

Axial Ratio Plot

1.26

3 dB Bandwidth= 31 MHz

Page 78: arm processor

Simple CMSA Modified CMSA

Simulation Results Comparison(Simple CMSA & Modified CMSA)

37.55 %

Antenna Efficiency Plot

36.99 %

Page 79: arm processor

Simple CMSA Modified CMSA

Simulation Results Comparison(Simple CMSA & Modified CMSA)

2.09

Antenna Gain Plot

2.02

Page 80: arm processor

Sr. No. Performance parameter Simple CMSA Modified CMSA

1 Bandwidth56 MHz (2.28%)

112 MHz (4.57%)

2Resonant Frequency 2.449 GHz 2.449 GHZ

3Axial Ratio 46.65 1.26

4Antenna Efficiency 37.55 % 36.99 %

5Gain 2.07 2.02

Simulation Results Comparison(Simple CMSA & Modified CMSA)

Page 81: arm processor

Experimental Results

Simple CMSA Modified CMSA

VSWR Plot

Bandwidth = 70 MHzfc = 2.47 GHz

Bandwidth= 140 MHzfc = 2.43 GHz

Page 82: arm processor

Experimental Results

Simple CMSA Modified CMSA

S11 Plot

Min. S11= -20 dB Min. S11 = -17.8 dB

Page 83: arm processor

Simulation and Experimental Result Comparison

Sr. No.

Performance Parameter

Simple CMSA Modified CMSASimulation

ResultExperimental

ResultSimulation

ResultExperimental

Result

1 Bandwidth (in MHz) 56 70 112 140

2 Bandwidth (in %) 2.28 2.83 4.57 5.76

3 Resonant Frequency (in GHz)

2.449 2.47 2.449 2.43

Page 84: arm processor

ConclusionBandwidth can be increased, almost double.Axial ratio is achieved near unity, so antenna can be said

as circularly polarized. No shift in resonant frequency of Modified CMSA

Slight reduction in antenna gain and efficiency.General model using ANN, for calculation of antenna

physical parameters is build and tested.Practical results are almost matching with simulated

results.

Page 85: arm processor

PublicationsSachin Takale, Dr. Shashikant Lokhande, “Design of Circular

Microstrip Antenna for Circular Polarization using ANN”, CiiT International Journal of Artificial Intelligence and Machine Learning, Nov. 2010, pp. 312-318 (Impact Factor – 0.765)

Sachin Takale, Dr. Shashikant Lokhande, “Design of single feed circularly polarized circular microstrip antenna”, National Conference on Pervasive Computing, NCPC-2010, April 9-10, 2010

Sachin Takale, Dr. Shashikant Lokhande, “Optimization of multilayered microstrip antenna using fuzzy-genetic approach”, e-PGCoN, April 28, 2008.

Page 86: arm processor

References[1] Constantine A. Balani,” Antenna Theory”, 3rd edition, John Wiley & Sons (Asia)

Pvt. Ltd. 2005. pp. 1-17,722-7846.

[2] Ramesh Garg, Prakash Bhartia, Inder Bahl, Apisak Ittipiboon, “Microstrip Antenna Design Handbook”, Artech House, Norwood, MA. 2001.

[3] Girish Kumar, K. P. Ray, “Broadband Microstrip Antennas” , Artech House, Norwood, MA. 2003

[4] S. N. Sivanandam, S. Sumathi, S. N. Deepa, “Introduction to Neural Networks using MATLAB 6.0”, Tata McGraw-Hill, 2006

[5] Howard Demuth, Mark Beale, “Neural Network Toolbox Version 4.0.4”, The MathWorks, Inc., Natick, MA, 2004.

[6] Constantine A. Balani,” Antenna Theory: A Review”, Invited Paper, Proceedings of the IEEE, vol. 80, no. 1, pp. 7-23, January 1992

[7] David M Pozar, “ Microstrip Antennas”, Invited Paper, Proceedings of the IEEE, vol. 80, no. 1, pp. 79-91, January 1992.

[8] Keith R. Carver, James W. Mink, “ Microstrip Antenna Technology”, IEEE transactions on Antennas and Propagation vol. AP-29, no. 1, pp. 2-23, January 1981.

Page 87: arm processor

References[9] S.Devi, Dhruba C Panda and Shyam S Pattnaik, “ A novel method of using

Artificial Neural Networks to calculate input impedance of circular microstrip antenna”, IEEE, 2002.

[10] E. R. Brinhole1, J. F. Z. Destro, A. A. C. de Freitas, and N. P. de Alcantara Jr.,“Determination of Resonant Frequencies of Triangular and Rectangular Microstrip Antennas, using Artificial Neural Networks”, Progress In Electromagnetics Research Symposium 2005, Hangzhou, China, August 22-26, pp. 579-582

[11] V. V. Thakare and P. Singhal, “ Bandwidth Analysis By Introducing Slots In Microstrip Antenna Design Using ANN, Progress In Electromagnetics Research M, Vol. 9, 107-122, 2009.

[12] V. V. Thakare and P. Singhal, “Neural network Based CAD model for the design of rectangular patch antennas”, Journal of Engineering and Technology Research Vol.1 (7), pp. 129-132, October, 2009

Page 88: arm processor

THANK YOU !!!!!