Skip to content

jeswinbenedict/Wireless-Computing-Non-CAT-and-Project

Repository files navigation

Network Intelligence Research Portfolio

A. Jeswin Karunya Benedict  ·  23MIC7225

School of Computer Science and Engineering  ·  VIT-AP University, Amaravati

Research IoT ML LaTeX


This repository contains academic research work exploring intelligent, adaptive, and secure network architectures — spanning vehicular ad hoc networks, IoT cybersecurity, and WSN-based routing systems.


Repository Structure

Network Intelligence Research
├── Assignment 1    —  Intelligent Routing Protocols for Smart Traffic
├── Assignment 2    —  Hybrid ML & DL Models for Robust IoT Cybersecurity
└── Research Project —  Adaptive Decision Support System for WSN-IoT Routing

Assignment 1 — Intelligent Routing Protocols for Smart Traffic and Urban Mobility Networks

TypeSurvey & Analytical Study
DomainVehicular Ad Hoc Networks (VANET) · Intelligent Transportation Systems (ITS)
ToolsLaTeX · draw.io · SUMO · OMNeT++ (referenced)

Overview

Modern urban environments face an exponentially growing traffic problem — not due to a lack of roads, but a lack of connected intelligence between vehicles. This paper examines how Intelligent Transportation Systems (ITS) leverage V2X communication and intelligent routing protocols to transform isolated vehicles into cooperative network nodes.

Key Protocol Families Analyzed

Protocol Family Latency Overhead Best Deployment Scenario
Proactive Topology-Based (DSDV, OLSR) Low High Stable RSU backhaul corridors
Reactive Topology-Based (AODV, DSR) High Low–Moderate Low-density environments (<40 vehicles/km)
Geographic / Position-Based (GPSR) Low Low–Moderate Dense urban grids — current best practice
Broadcast / Geocast Very Low High Safety-critical message dissemination
Learning-Based Adaptive (DRL, Federated RL) Low–Moderate Moderate Near-future adaptive urban deployment

Key Insights

  • Road-aware geographic forwarding outperforms map-oblivious protocols by 30–40% PDR improvement in organic urban topologies (European-style cities)
  • Reactive protocols become obsolete beyond approximately 50 vehicles/km due to self-interfering RREQ floods
  • Deep Reinforcement Learning routing achieves near-optimal delivery rates across all density profiles — the primary challenge being simulation-to-real-world transfer
  • Federated learning is identified as the bridge to scalable, privacy-preserving on-road RL deployment

Architecture Covered

  • V2V, V2I, V2P, V2N communication modalities
  • DSRC (IEEE 802.11p @ 5.9 GHz) vs. Cellular V2X (C-V2X / 3GPP Rel. 14+)
  • RSU gateway integration with cloud traffic management backends
  • SUMO + OMNeT++ (Veins/TraCI) simulation methodology for realistic mobility modeling

Assignment 2 — Hybrid Machine Learning and Deep Learning Models for Robust IoT Cybersecurity

TypeTechnical Survey & Model Analysis
DomainIoT Security · Intrusion Detection · Threat Intelligence
ApproachHybrid ML/DL fusion for multi-vector cyberattack defense

Overview

With billions of IoT devices deployed globally across smart cities, healthcare, and industrial systems, the attack surface is unprecedented. This work investigates how hybrid combinations of classical machine learning and deep learning architectures can produce more robust, generalizable intrusion detection and anomaly detection systems for IoT environments.

Core Focus Areas

  • Threat Landscape — DDoS, Man-in-the-Middle, Replay Attacks, Botnets targeting IoT
  • Classical ML Approaches — Random Forest, SVM, XGBoost for feature-engineered detection
  • Deep Learning Models — LSTM, CNN, Autoencoders for sequential and raw-traffic analysis
  • Hybrid Architectures — Ensemble and multi-stage fusion models for improved precision and recall
  • Federated Security — Distributed model training across edge IoT nodes without raw data exposure

Key Themes

  • Trade-offs between detection accuracy, inference latency, and resource constraints on edge devices
  • Handling class imbalance in IoT attack datasets (NSL-KDD, UNSW-NB15, Bot-IoT)
  • Transfer learning and domain adaptation for cross-platform generalization

Research Project — Adaptive Decision Support System Based Heterogeneous Routing Protocol for WSN-Based IoT Network

TypeOriginal Research Paper
Venue TargetElsevier CAS Double-Column Journal Format
DomainWireless Sensor Networks · IoT Routing · Adaptive Systems
AffiliationVIT-AP University, Amaravati, India

Abstract

WSN-IoT integration suffers from the continuous problem of routing as the environment poses issues such as heterogeneous devices, dynamic routing topologies, and extreme power limitations. Traditional routing techniques target single optimization criteria and hence are unable to handle situations involving multi-criteria such as power, reliability, latency, and security in the environment. This paper presents a new adaptive decision support framework handling heterogeneous routing via multicriteria optimization based on residual energy, link quality, node heterogeneity factors, load conditions, and risks of attacks.

Research Objectives

  1. Integrated Routing Metric Framework — Unify residual energy, link quality, node heterogeneity, traffic congestion, and security risk into a single multi-criteria decision model
  2. Lightweight MCDM Implementation — Design an AHP/TOPSIS-based decision engine fit for resource-constrained WSN nodes with minimal compute and memory overhead
  3. Predictive ML Routing — Deploy ML classifiers/regressors to predict link degradation, congestion, and node failure proactively — reducing reactive route repairs and control overhead
  4. Simulation Validation — Benchmark ADSS against AODV, RPL, LEACH, and bio-inspired protocols across dense urban and sparse industrial topologies

Proposed System: ADSS — Adaptive Decision Support System

+------------------------------------------------------------------+
|                         ADSS Framework                          |
|                                                                  |
|  +--------------+    +---------------+    +-----------------+   |
|  |    Metric    |    |  MCDM Engine  |    |   Predictive    |   |
|  |  Collector   |--->|  (AHP/TOPSIS) |--->|   ML Module     |   |
|  |              |    |               |    |  (Link/Energy)  |   |
|  +--------------+    +---------------+    +--------+--------+   |
|                                                    |            |
|  Inputs: Residual Energy · Link Quality ·          v            |
|  Node Heterogeneity · Traffic Load ·       +---------------+    |
|  Attack Risk Score                         |    Adaptive   |    |
|                                            | Route Decision|    |
|                                            +---------------+    |
+------------------------------------------------------------------+

Simulation Results Overview

The following performance metrics were evaluated across varying node densities and traffic conditions:

Metric ADSS AODV RPL LEACH
Packet Delivery Ratio (PDR) Superior Baseline Moderate Moderate
Network Lifetime Extended Moderate Moderate Limited
Energy Efficiency Optimized Average Average Cluster-limited
Routing Overhead Controlled High (reactive) Moderate Low
Attack Resilience (PDR under attack) Robust Vulnerable Moderate Vulnerable

Results generated via NS-2/NS-3 compatible simulation scripts (.cc source files included)

Project File Structure

Research Project
├── main.tex                      # Full paper in Elsevier CAS format
├── references.bib                # 30+ curated references (post-2023 journals)
├── source/
│   ├── adss_wsn.cc               # Core ADSS protocol implementation
│   ├── adss_pdr_comparison.cc    # PDR benchmarking script
│   ├── adss_qos_analysis.cc      # QoS evaluation
│   ├── adss_lifetime_analysis.cc
│   ├── adss_energy_overhead.cc
│   ├── graph01-graph12.tex       # PGFPlots graph sources
│   └── Fig1.drawio / fig2.drawio
├── results/
│   ├── graph00_comparative_summary_panel.png
│   ├── graph01_pdr_vs_nodes.{csv,png}
│   ├── graph02_lifetime_vs_nodes.{csv,png}
│   ├── graph03_delay_vs_nodes.{csv,png}
│   ├── graph04_energy_vs_nodes.{csv,png}
│   ├── graph05_pdr_vs_traffic.{csv,png}
│   ├── graph06_lifetime_vs_energy.{csv,png}
│   ├── graph07_weight_evolution.{csv,png}
│   ├── graph08_throughput_vs_nodes.{csv,png}
│   ├── graph09_pdr_vs_attack.{csv,png}
│   ├── graph10_overhead_vs_nodes.{csv,png}
│   ├── graph11_survival_vs_time.{csv,png}
│   └── graph12_delay_vs_traffic.{csv,png}
├── figures/
│   ├── figure1.png               # System architecture diagram
│   └── figure2.png               # Protocol flow illustration
└── PAPERS/
    ├── B1-B8                     # Base papers (journals, post-2023)
    └── R1-R26                    # Reference papers (30+ curated sources)

Technologies and Tools

Category Tools
Writing & Typesetting LaTeX · Overleaf · Elsevier CAS Template · BibTeX
Simulation NS-2/NS-3 · SUMO · OMNeT++ (Veins/TraCI)
Diagramming draw.io · PGFPlots / TikZ
Languages C++ (simulation scripts) · LaTeX
Domains VANET · WSN · IoT · Cybersecurity · Adaptive Routing

Research Themes

Theme Keywords
Intelligent Routing VANET · Geographic Forwarding · DRL-based Routing
WSN-IoT Optimization Energy Efficiency · Network Lifetime · Multi-hop Routing
Adaptive Decision Systems AHP · TOPSIS · MCDM · Predictive ML
IoT Cybersecurity Intrusion Detection · Hybrid ML/DL · Federated Security
Network Resilience Attack-aware Routing · Fault Tolerance · QoS Guarantees

Author

A. Jeswin Karunya Benedict  ·  23MIC7225

School of Computer Science and Engineering VIT-AP University · Amaravati, Andhra Pradesh, India

jeswin.23mic7225@vitapstudent.ac.in  ·  ORCID: 0009-0004-2202-9012

Advisor: S. Gopikrishnan  ·  gopikrishnanme@gmail.com


© 2025 A. Jeswin Karunya Benedict · VIT-AP University · All research content is for academic purposes.

About

Academic research on intelligent routing protocols for VANETs, hybrid ML/DL models for IoT cybersecurity, and an adaptive decision support system for WSN-IoT routing optimization.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors