Energy-Efficient Optimization of Heterogeneous Wireless Sensor Networks Using a Two-Phase Genetic Algorithm

Wireless Sensor Networks (WSNs) are used everywhere today — in agriculture, traffic systems, environmental monitoring, smart cities, and even wildlife tracking. They collect real-time data through small sensor nodes spread across large areas.

But they have one huge problem:

⚠️ The Biggest Enemy of WSNs: Battery Life

Most sensor nodes run on tiny batteries. They are placed in forests, mountains, buildings and remote areas. Changing or recharging them is often impossible.

So if one important node dies early, the entire network becomes weak. If many nodes die early, the network collapses.

This is even more critical in Heterogeneous Wireless Sensor Networks (HWSNs) — where we have:

• Normal nodes → low energy
• Super nodes → high energy (they act as cluster heads)

But even super nodes drain out quickly because they have the heavy responsibility of:

• Collecting data from normal nodes
• Processing the data
• Sending it to the Base Station (BS)

So the main question is:

❓ "How do we make the network survive longer while using the same battery?"

This is the exact problem the research paper aimed to solve.

---

🧠 The Core Idea of the Research

The research introduces a Two-Phase Genetic Algorithm (GA) approach to improve:

✔️ Sleep scheduling

Deciding which super nodes should stay awake and which should rest (sleep mode).

✔️ Routing

Building the best multi-hop path so data travels efficiently.

✔️ Clustering

Grouping normal nodes under the best possible super nodes (cluster heads).

By combining all three together, the algorithm creates a more intelligent network that saves energy and survives much longer.

---

🧩 Phase 1: Smart Sleep Scheduling + Routing Tree

The algorithm first decides:

• Which super nodes must stay awake (others sleep)
• How they connect with each other to build a routing tree

✨ Unique idea:
The network is divided into rings around the Base Station.
Each ring wakes up an equal number of super nodes.

This ensures:

• Coverage is uniform
• No region becomes "empty"
• No super node is overloaded

---

🧩 Phase 2: Clustering Normal Nodes

After deciding awake super nodes and building a routing tree, the algorithm assigns normal nodes to the best possible cluster heads.

It tries to:

• Distribute normal nodes evenly
• Avoid overloading any single super node
• Reduce communication distance
• Save energy during data transmission

---

⚙️ How the GA Solves the Problem

The GA uses:

• Custom chromosome structure
• Specialized selection, crossover and mutation
• Energy-aware cost functions
• Repair operations to maintain valid trees
• Probabilistic selection based on remaining energy

---

📊 Results of the Proposed Method

1️⃣ Improved Cluster Stability & Coverage

Coverage remains consistently higher than competing algorithms.

---

2️⃣ Balanced Cluster Sizes

Cluster sizes remain stable, consistent and efficient.

---

3️⃣ Better GA Cost Optimization

Phase 1 and Phase 2 GA cost improvements can be seen clearly.

---

4️⃣ Final Cluster Distribution

The distribution of cluster sizes at the final round indicates healthy balance.

---

🧑‍💻 Project Source Code (GitHub Repository)

If you want to explore the full implementation, experiment with parameters, or reproduce all the graphs:

📦 Download Source Code

This repository includes:

• Complete Python code
• Genetic Algorithm implementation
• Sleep scheduling logic
• Clustering logic
• Routing tree generation
• All plotting scripts
• A clear README

---

🧾 What Problem Did We Actually Solve?

This project solves three major problems in WSNs:

✔️ Unbalanced energy consumption

Some super nodes earlier used to die faster.
Now energy load is spread evenly.

✔️ Poor coverage

Sleeping nodes sometimes left regions uncovered.
Now every ring guarantees awake nodes.

✔️ Short network lifetime

The network usually died early due to poor scheduling.
With the new method, lifetime extends by hundreds of rounds.