🛰️ NOMA + SIC Implementation

🔸 What is NOMA?

NOMA (Non-Orthogonal Multiple Access) allows multiple devices (drones) to transmit over the same frequency band simultaneously by allocating different power levels to each device. The base station decodes signals using Successive Interference Cancellation (SIC).

🔸 How We Implemented NOMA

In our simulator (comm/noma.py):

Power Mapping
Each drone is assigned a predefined power level:

NOMA_POWER_LEVELS = [5, 4, 3, 2, 1]  # Drone 0 has highest decoding priority

Superposition of Signals
Encrypted drone messages are converted to byte arrays, scaled by their NOMA power level, and numerically added:

total += arr * self.power_map[i]  # Simulate analog superposition

Raw Buffers Stored
Individual encrypted buffers are stored for later decoding.

SIC-Like Decoding
Base station decodes messages in descending power order:

for i, _ in sorted(self.power_map.items(), key=lambda kv: -kv[1]):
    decrypted = ChaCha20.decrypt(raw[i])
    decoded[i] = decrypted

✔️ While not RF-accurate, this provides strong conceptual simulation[8].

---

🔐 ChaCha20 Encryption Integration

🔸 What is ChaCha20?

ChaCha20 is a stream cipher offering high-speed encryption and modern cryptographic security, ideal for resource-constrained systems like UAVs.

🔸 How We Use It

In comm/encryption.py, every message:

1. Is serialized with json.dumps(...) and encrypted:

ciphertext = cipher.encrypt(plaintext)

2. Is decrypted symmetrically at receiver with same key + nonce.
   Both drones and base station use shared CHACHA20_KEY and CHACHA20_NONCE.

Secure Communication Flow

Drone State → [Encrypt with ChaCha20] → NOMA superposition → Base Station
Base Command → [Encrypt with ChaCha20] → TCP → Drone

---

🚧 Obstacle Avoidance Logic

How Drones Avoid Collisions:
Each drone (drone_agent.py) uses:

360° Vision Radius

VISION_RADIUS = 50

Obstacle Sensing
Checks all obstacles within vision radius:

if math.hypot(self.position[0]-ox, self.position[1]-oy) <= VISION_RADIUS:

Reactive Avoidance
Computes repulsion vector when obstacle detected:

dx, dy = x - ox, y - oy
push = 5
self.position = (x + dx / dist * push, y + dy / dist * push)

Enables local real-time collision mitigation.

---

🧠 Area Optimization (Coverage)

User Coverage Optimization (in base_station_server.py):

1. Users spawned randomly in world.py
2. Each drone:
   • Checks if user within its VISION_RADIUS
   • Reports to base station via NOMA
3. Base Station:
   • Tracks covered users
   • Sends target positions to drones:
     • Directs drones toward nearest uncovered users (Euclidean distance)
     • If all users covered, drones patrol randomly

def nearest_uncovered_user(drone_pos, users, covered_users):
    # Picks nearest user not already covered

Ensures maximized coverage[7].

---

🛰️ System Workflow Summary

┌────────────┐       NOMA+ChaCha       ┌──────────────┐
│  Drone #i  │ ────────────────────▶   │ Base Station │
│ [Obstacle] │        Uplink           │  (Autonomy)  │
└────────────┘ ◀────────────────────   │   Controller │
      ▲             Encrypted          └─────┬────────┘
      │                                      │
      ▼                                      ▼
Local avoidance              Selects new target (user, patrol)
via obstacle sense           Sends `move` command (ChaCha)

---

🛠️ Customization Capabilities

Feature	How to Configure
Drone Count	NUM_DRONES in config.py
Vision Radius	VISION_RADIUS in config.py
World Area	WORLD_LIMIT in config.py
Users + Obstacles	World._spawn_users/obstacles()
Ping Interval	Change timer in base_station_server.py
NOMA Powers	Modify NOMA_POWER_LEVELS

---

🔮 Future Scope

• Train RL agents using simulator by plugging into DroneAgent movement policy
• Integrate MADDPG/PPO with realistic communications
• Add dynamic user mobility, battery constraints, dropout scenarios
• Log telemetry in JSONL/CSV for analysis
• Add wireless physics (fading, interference) using ns3/GNURadio

---

Made with Python🐍 arjun7579