Overview on AgriFoRwArdS Summer School outcomes Group "Baa‐ndwidth" - LCAS/auto_shepherd_communications GitHub Wiki

Baa-ndwidth Team: Communications Systems Research Report

Group 2: Autonomous Shepherd Communications

Team Members:

  • Callum Lennox
  • Dimitris Paparas
  • Robbie Cato
  • Liyou Zhou
  • Robert Stevenson
  • James R. Heselden

Slides: Baa-ndwidth_Summer_School_2025.pdf

Project Overview

The Baa-ndwidth Team was tasked with establishing and encouraging data communications between groups for autonomous shepherd systems, including evaluation of communication hardware and protocols. The project aimed to develop infrastructure-free communications solutions capable of connecting robots in field environments.

Project Objectives

Primary Goals

  • ROS2 Group Messages: Establish protocols for inter-group message sharing and construct a shared ROS2 messages package
  • ROS2-LoRa Bridge & ROS2-nRF Bridge: Configure serial-to-radio protocols for long-range infrastructure-free communications
  • Empirical RF Propagation Study: Perform radio frequency survey across Riseholme Campus, mapping signal attenuation

Stretch Objectives

  • Deploy radio communication package on field robots
  • Conduct viability assessments for publication
  • Develop mesh infrastructure deployment capabilities

Communication Protocols Evaluated

1. Long Range Wi-Fi (IEEE 802.11LR)

Technical Specifications:

  • Standard: IEEE 802.11LR
  • Technology: Reuses existing 802.11b physical layer with narrowed data rate
  • Performance Improvement: 4 dB improvement in reception sensitivity vs 802.11b
  • Hardware Support: Supported natively by ESP32
  • Antenna: PCB integrated antenna
  • Range Achieved: 130 metres with minimal latency (PCB antenna)
  • Potential Range: Up to 300 metres with external antenna

Performance:

  • ✅ Excellent for standard Wi-Fi applications
  • ✅ Good latency characteristics
  • ✅ Native ESP32 support
  • ✅ Improved sensitivity over standard 802.11b
  • ❌ Limited range compared to LoRa solutions
  • ❌ Requires infrastructure or access points for full Wi-Fi stack

2. LoRa (Long Range Radio)

Technical Details:

  • Standard: IEEE 1451.0 (TEDS)
  • Frequency: 868 MHz (UK/Europe), 433 MHz also tested
  • Data Rate: 0.3 kbit/s to 50 kbit/s per channel
  • Link Budget: 155 dB - 170 dB
  • Modulation: Long Range Frequency Hopping Spread Spectrum (LR-FHSS)
  • Layer: Physical layer (Layer 2) radio-based protocol
  • Theoretical Range: Up to 1 kilometre in open space line-of-sight
  • Coding Rate: 4:5 (for every 4 bits sent, 1 additional integrity bit generated)
  • Channels: 8 channels available
  • Bandwidth: 120 Hz over 868 MHz broadcast frequency
  • Packet Structure: Sync word, header (metadata), payload with CRC

Performance Results:

  • Initial Setup Time: ~1 day to optimise speed
  • Round-trip Time: Improved from 550ms to 22ms (sending Twist message + acknowledgement)
  • Packet Integrity: Excellent due to forward error correction
  • Range: Excellent penetration, only lost connection twice in deep woods
  • Connection Reliability: Very good overall

Key Limitations:

  • Duty Cycle Restrictions: 1% duty cycle on public gateways in UK (36 seconds airtime per hour)
  • Legal Compliance: Must adhere to Ofcom guidelines (community-policed)
  • Real-time Control: Not suitable for frequent back-and-forth communication
  • Listen Before Talk: Additional protocol requirements

Regulatory Considerations:

  • UK: 1% duty cycle standard, up to 10% on certain bands (360 seconds per hour)
  • Community self-regulation (e.g., Meshtastic firmware locks to prevent violations)
  • Different restrictions globally

3. ESP-NOW

Technical Specifications:

  • Frequency: 2.4 GHz
  • Protocol: Proprietary Wi-Fi stack compression by Espressif
  • Architecture: Wireless communication protocol based on data-link layer
  • OSI Model: Reduces five layers of OSI model to only one layer
  • Layer Bypass: Data doesn't need to pass through network, transport, session, presentation, or application layers
  • Packet Headers: No need for packet headers or unpackers on each layer
  • Maximum Payload: 250 bytes officially (theoretical maximum: 2,332 bytes with modifications)
  • Communication: Bidirectional peer-to-peer (single-duplex or full-duplex)
  • Encryption: Supported
  • Infrastructure: No external Wi-Fi source or router required
  • Setup Complexity: Very simple

Key Advantages:

  • ✅ Minimal overhead and excellent latency due to simplified stack
  • ✅ Quick response, reducing delay caused by packet loss in congested networks
  • ✅ Surprisingly decent range for simplicity
  • ✅ No infrastructure required (peer-to-peer)
  • ✅ Easy mesh network implementation
  • ✅ Direct MAC address communication
  • ✅ Ubiquitous ESP32 hardware support

Key Discovery:

  • MAC addresses are editable despite multiple sources (including Espressif documentation) stating they're "burned in at factory"
  • This caused debugging issues initially but provides valuable flexibility for network management

4. nRF24L01 (Nordic Radio)

Technical Specifications:

  • Manufacturer: Nordic Semiconductors
  • Frequency Range: 2.4 - 2.5 GHz ISM band
  • Data Rates: 250 Kbps / 1 Mbps / 2 Mbps
  • Channels: 125 different channels (1 MHz spacing, 2400-2525 MHz)
  • Network Capability: 125 independently working modems in one location
  • Addressing: Up to 6 addresses per channel, or each unit can communicate with up to 6 other units simultaneously
  • Range Enhancement: External antenna variation includes RFX2401C chip with PA (Power Amplifier) and LNA (Low-Noise Amplifier)
  • Potential Range: Up to 1000 metres in open space with external antenna

Status: ❌ Problematic Implementation

Issues Encountered:

  • Should theoretically work with common setup
  • Experienced persistent "gremlins" and connectivity issues
  • Took a week of troubleshooting with no successful operation
  • Previously worked in other contexts, suggesting environmental factors

Suspected Causes:

  • Poor physical wires and connectors
  • Radio interference from multiple devices in shared workspace
  • Wi-Fi interference from concurrent testing (both operate at 2.4 GHz)
  • Hardware compatibility issues

ROS2 Integration

Communication Protocol Design

Message Structure:

[2 bytes]  - Message start marker
[4 bytes]  - Payload length
[1 byte]   - ROS2 message type
[Variable] - Serialised ROS2 payload (using rclpy.serialization.serialize_message())

Supported Message Types:

  • PoseStamped: Robot positioning commands with header (timestamp, frame_id) and pose (position: x,y,z + orientation: x,y,z,w)
  • GPS: Location feedback from robots with header (timestamp, frame_id), status, service, latitude, longitude, altitude, and position covariance

Implementation:

  • Bidirectional communication achieved
  • Successful message serialisation and deserialisation using ROS2 native functions
  • Terminal-based testing with publish/subscribe pattern
  • Protocol tested using ESP-NOW transport layer
  • Integration designed for robot dog platforms

Empirical RF Propagation Study

Methodology

Test Setup:

  • Base station positioned behind Riseholme Old Hall (on lawn between building and lake)
  • Mobile receiver attached to tester's head
  • Walking survey conducted across extended campus
  • Signal strength (RSSI) mapping performed

Results

Results

LoRa Performance Map

  • RSSI Range: -40 dBm to -140 dBm (colour-coded heat map)
  • Coverage: Maintained connectivity across most of campus
  • Signal Loss: Occurred when line-of-sight blocked by multiple trees and dense foliage
  • Penetration: Generally superior range and penetration through obstacles
  • Consistency: Reliable signal maintained throughout most of the walking survey

ESP-NOW (Wi-Fi with External Antenna) Performance Map

  • RSSI Range: -40 dBm to -140 dBm (colour-coded heat map)
  • Coverage: More susceptible to foliage interference than LoRa
  • Signal Loss: Clear line-of-sight dependency, particularly affected by tree coverage
  • Degradation: Faster signal degradation at higher RSSI values
  • Error Correction: Less effective error correction compared to LoRa
  • Pattern: Shows distinct blue areas (no signal) where LoRa maintained connectivity

Visual Evidence: The heat maps clearly demonstrate LoRa's superior penetration through the wooded areas around Riseholme Lake, whilst ESP-NOW shows more dramatic signal dropouts in areas with dense tree coverage.

Important Caveats

Test Limitations:

  • Different antennas used for LoRa vs ESP-NOW
  • Different frequencies tested (868 MHz vs 2.4 GHz)
  • Antenna gain values potentially different
  • Not true apples-to-apples comparison

Lessons Learned

Technical Insights

  1. Long Range Wi-Fi (IEEE 802.11LR): Provides 4 dB sensitivity improvement over standard 802.11b through data rate narrowing, offering enhanced range whilst maintaining Wi-Fi compatibility
  2. LoRa: Excellent for low-frequency sensor data with superior obstacle penetration, but unsuitable for real-time robot control due to duty cycle limitations (36 seconds per hour at 1% duty cycle)
  3. ESP-NOW: Outstanding choice for robotics applications requiring frequent communication within moderate ranges, with simplified single-layer architecture providing exceptional latency performance
  4. nRF24: Potentially viable but requires careful environmental consideration and quality components to avoid interference issues
  5. Antenna Design: External antennas dramatically improve range for all protocols tested

Practical Considerations

  1. Regulatory Compliance: Always verify duty cycle restrictions and frequency allocations for your region
  2. Environmental Testing: Real-world RF propagation differs significantly from theoretical calculations
  3. Protocol Selection: Match communication requirements to protocol capabilities:
    • High-frequency control: ESP-NOW or Wi-Fi
    • Occasional sensor data: LoRa
    • Infrastructure available: Wi-Fi
    • Mesh networking: ESP-NOW

Development Recommendations

  1. Quality Components: Invest in reliable connectors and wiring
  2. Environmental Isolation: Test radio systems away from interference sources when possible
  3. MAC Address Management: Document and plan ESP-NOW device addressing carefully
  4. Range Testing: Conduct thorough field trials in actual deployment environment

Hardware Recommendations

Successful Platforms

  • Waveshare RP2040 with LoRa SX1276: Reliable for LoRa applications
  • ESP32 development boards: Excellent for ESP-NOW implementations
  • External antennas: Significant improvement over PCB antennas

Libraries and Tools

  • RadioLib: Comprehensive radio communication library
  • Arduino IDE: Primary development environment
  • ROS2: Message serialisation and communication framework

Future Directions

Immediate Applications

  • Field robot deployment with ESP-NOW for control systems
  • LoRa sensor networks for environmental monitoring
  • Mesh networking for multi-robot coordination

Research Opportunities

  • Mesh infrastructure deployment across multiple sites
  • Integration with existing robotic platforms
  • Performance optimisation for specific use cases
  • Publication of propagation study results

Resources and Links

Project Repositories

Video Documentation

Educational Resources

Technical References

Conclusion

The Baa-ndwidth Team successfully evaluated multiple communication protocols for autonomous shepherd applications, conducting comprehensive technical analysis and real-world testing. ESP-NOW emerged as the optimal solution for real-time robot control, offering excellent range, low latency through its simplified single-layer architecture, and straightforward implementation. LoRa proved ideal for sensor networks where occasional data transmission is sufficient, demonstrating superior penetration through obstacles but being limited by regulatory duty cycle restrictions.

The empirical RF propagation study provided valuable visual evidence through RSSI heat maps, clearly demonstrating the different propagation characteristics of each protocol across varied terrain. The project's ROS2 integration work established practical implementation pathways using proper message serialisation protocols.

Key technical insights include the discovery that ESP32 MAC addresses are indeed editable (contrary to common documentation), the importance of antenna selection for range performance, and the critical impact of environmental factors on 2.4 GHz vs sub-GHz performance.

The project establishes a foundation for robust field communication systems and provides clear guidance for protocol selection based on specific application requirements. Future work should focus on mesh network deployment and integration with autonomous robotic platforms.

"Thanks for listening to our presentation! Please don't ask us any questions..." - Baa-ndwidth Team

Report compiled using LLM from presentation transcript and slide documentation