AEP-DNN BEAM STEERING SYSTEM

NeuroBeam

INITIALIZING ARRAY // 28GHZ // Ka-BAND

5G Ship-to-Satellite Communication

TOUCH ANYWHERE TO START

Final Year Project — 28 GHz 5G Communication

NeuroBeam Beam Steering

Deep Neural Network for real-time phased array beam steering.
Enabling reliable ship-to-satellite communication at millimeter-wave.

1.38° Phase MAE
<1ms Prediction
90% Link Availability
28 GHz Ka-Band
Try Live Data ↓

See NeuroBeam In Action

Watch how our DNN-powered beam steering system works — from MATLAB simulation to real hardware deployment on TMYTEK BBox.

▶ CLICK TO INTERACT

* In video data are dummy,scroll down and contact us for real data

01
MATLAB
Simulation
02
DNN
Training
03
Real-Time
Prediction
04
Hardware
Deployment
05
Auto
Tracking
MATLAB Phased Array Simulation
Generate training data by simulating a 4-element ULA with mutual coupling at 28 GHz. 121 steering angles across ±60° produce the phase shift dataset that trains the DNN.

How It Works

01

Data Generation

Simulate 4-element ULA with mutual coupling at 28 GHz using MATLAB Phased Array Toolbox. Generate 121 training samples across ±60°.

AEP_DNN_DataGeneration.m
02

DNN Training

Train custom neural network (1→64→128→64→8) with sin/cos phase encoding. Backpropagation from scratch, no Deep Learning Toolbox needed.

AEP_DNN_Training.m
03

Validation

Three-way comparison: Mathematical model vs DNN vs Optimal. DNN achieves 1.38° MAE with mutual coupling compensation. Verified on TMYTEK TMXLABKIT hardware.

AEP_DNN_Validation.m
04

Real-Time Prediction

Predict phase shifts, voltages, and power in <1ms. Hardware-ready outputs for 6-bit phase shifter with 0–5V control.

predictWithPolarPattern.m
05

Auto-Tracking

Closed-loop Python script connects to TMYTEK BBox via Ethernet. Automatically sweeps beam to lock onto the strongest LTC5596 power signal.

auto_track_bbox.py

DNN Architecture

Input
1 neuron
Target Angle
Hidden 1
64 neurons
tanh
Hidden 2
128 neurons
tanh
Hidden 3
64 neurons
tanh
Output
8 neurons
4 sin + 4 cos
Parameters ~17,000
Training Time 18.3s
Best Epoch 2,432
Encoding Sin/Cos

Phase Prediction Data

Adjust the steering angle to see predicted phase shifts, voltages, and radiation pattern for the 4-element array. Power values are from real TMYTEK hardware measurements.

-60°-30°+30°+60°
🔒
Unlock Hardware Data
ElementPhaseStepVoltage

Live Hardware Control

Connect to TMYTEK BBox N257 via the local bridge server. Control phase shifts and perform automatic beam tracking in real-time.

Disconnected
OFFLINE
+0°
-60°+60°
Received Power (LTC5596)
-- dBm
📡 System Log ● LIVE
Waiting for bridge connection...
⚙️
Hardware Bridge Not Running
Start the bridge server on your PC:
python hardware_bridge.py --simulate
The panel will auto-connect when the bridge is detected.

Ship-to-Satellite Results

📡

SNR Improvement

+15–20 dB

Over fixed antenna (no beam steering)

🔗

Link Availability

~90%

vs 50% without steering during LEO pass

📶

Channel Capacity

2.5×

Improvement over no-steering baseline

🎯

Beam Accuracy

<1°

Pointing error despite ship roll ±8°

Three-Scenario Comparison

Metric No Steering Mathematical AEP-DNN
Link Availability ~50% ~75% ~90%
Avg Channel Capacity ~100 Mbps ~200 Mbps ~250 Mbps
Pointing Error Large ~5° <1°
Coupling Compensated No No Yes

System Specifications

Antenna Array

  • Type4-Element ULA
  • Frequency28 GHz (Ka-band)
  • Wavelength10.71 mm
  • Element Spacing5.36 mm (0.5λ)
  • Steering Range±60°

Phase Shifter

  • HardwareTMYTEK TMXLABKIT
  • DetectorLTC5596 (100MHz–40GHz)
  • Resolution6-bit (64 steps)
  • Step Size5.625°
  • Voltage0–5 V
  • Tx Power20 dBm/element

Hardware Bridge

  • LanguagePython 3
  • FrameworkFlask REST API
  • TMYTEK Librarytlkcore
  • BBox InterfaceEthernet (TCP/IP)
  • LTC5596 InterfaceUSB Serial
Project Developer: Nazrul Alzam Bin Mat Akher (ID: 221022252) | nazrul.alzam14@gmail.com | +60174933296