In this C++ project, I created a TCP server capable of detecting anomalies in a time-series data file. It does so by first learning what a "normal" data set is considered to be, and then analyzing any data set with the same features and calling out anomalies in it.
This project shows the use of:
- The Command and Bridge design patterns.
- Server-Client communication over the TCP protocol.
'AnomalyDetector.h' - declares an interface for TimeSeries anomaly detectors as well as 2 ways of representing anomalies.
SimpleAnomalyDetector.h - declares a specific anomaly detector implementing the AnomalyDetector interface. It uses a linear regression algorithm for the anomaly detection.
SimpleAnomalyDetector.cpp - the main unit of the project. It studies the correlations in a 'normal' data-set, and then searches for anomalies in any data-set, based on the knowledge gained during the learning phase.
HybridAnomalyDetector.h - declares another specific anomaly detector. It uses both linear regression and minimal encolsing circle algorithms for the anomaly detection.
HybridAnomalyDetector.cpp - the hybrid anomaly detector inherits from SimpleAnomalyDetector and overrides its relevant methods. it utilizes dynamic binding of class methods to insure the override is successful during run-time.
CLI.cpp - implements the CLI for the server-client communication by using an array of Commands.
Server.h - declares the Server and ClientHandler classes. Implements the Brigde design pattern.
Server.cpp - creates and runs a TCP server.
anomaly_detection_util.cpp - includes the computational part of the linear regression algorithm, bases on statistic linear regression methods using pearson's correlation. An example for the output would be: [line # col1-col2]
commands.h - declares and implements a set of commands to be used in the CLI. implements the Command design pattren.
minCircle.cpp - impelements Welzl's algorithm for finding the minimal enclosing circle of a set of points in a 2D plane.
timeSeries.cpp - a class representing the time-series data. It extracts the feature vectors from a columb-based CSV file.
MainTrain.cpp - a main function to run the program, along with a test of 2 clients.\