|
 Variable User Mobility Analysis using Stochastic Directional
Correlation Concept in Mobile Network System
Abstract
Stochastic nature of variable
user mobility is one of the basic fields of research for traffic
performance analysis for a given mobile network system. Due to heavy
burden of space and time complexity in computation that arises during
simulation when significant mobility parameters are considered, several
analytical models have been proposed as alternatives. Moreover the analytical approach
to figure out the actual stochastic behavior is so difficult that exact
realistic features can hardly be implemented in the model without
simplification. Hence the effect of variable user mobility in the traffic
performance may not be analyzed with such simplified and inappropriate
model. This paper has focused on the developing the variable user
mobility model with the effect of stochastic direction correlation
concept in the hexagonal cell structured network. The concept of velocity
transition matrix is used to differentiate the effect of different users
in the network.
Identification of clusters in an Object-Oriented Software Design
using Graph Spectra
Abstract
This
paper proposes a method for identification of densely coupled communities
of classes (clusters) in object oriented software system designs. These
clusters decompose the software system into smaller subsystems which are
relevant in terms of functionality. Utilizing this, modules can also be
obtained that form the system. This method also identifies autonomous
clusters and implies reusable components. The term “reusable
components” is used to refer to the set of interrelated classes
representing a module to be used in different software systems. Finally
this method estimates the degree of modularity in an object oriented
software design. To accomplish this we employ graph partitioning using
its spectra. The class diagram of the software system is modified into a
graph with classes represented by nodes and the interrelationships
between classes determining the weights of the edges. Our approach is
based on a recursive partitioning using the spectra of the Laplacian
matrix of the graph. Emphasis has been laid out in considering the
Fiedler value for partitioning of the graph.
Stock Price Movement Analysis using
Fuzzy Logic and Technical Analysis
Abstract
The application of technical analysis in the prediction of future
stock price movement overrules certain factors such as government’s
fiscal policy, market trends, economic environment and political issues.
The only concern in technical analysis is the past movement of stock prices,
preferably that of recent past and the forces of supply and demand that
affect those prices. The proposed method incorporates such input
parameters discarded in technical analysis into the fuzzy logic system. Fuzzy
informational decision making is applied to investment analysis through
technical analysis. Fuzzy logic has been used quite extensively along
with technical analysis. But an independently functioning fuzzy real time
system has not been developed. The proposed model is an independently
functioning real time model which predicts future stock values. The model
is based on the previously proposed models, improvises over indicators
proposed in earlier models and introduces interdependencies among
indicators. These interdependencies accounts for the recent changes
observed during stock price evaluation. The simulation of human behavior
is incorporated in the analysis of stock price movement. The method is
dependent on fuzzy logic for decision making when certain price movements
are observed. The performance and success of the model is measured in
terms of the difference between predicted and actual stock price
movement. The proposed stock price prediction model has been shown to
exceed investment returns. The flexibility and versatility of the system
is also demonstrated.
A Fuzzy Logic Based approach for detection of significant
vertices for polygonal approximation of digital curves
To be published later
|
