LoRa Mesh Details

Meshtastic wire format, channels, and flood routing on GhostESP

GhostESP speaks the stock Meshtastic wire format so it interoperates with unmodified firmware.

Wire format

Each LoRa frame is:

[16 B header] [AES-CTR ciphertext: protobuf Data]
  • Header (16 B): from, to, packet ID, flags, channel hash, next hop, and relay node.
  • Ciphertext: AES-CTR with the channel PSK (LongFast default PSK AQ== → chash 0x08) covering the serialized Data protobuf (portnum, payload, routing fields).
  • De-dup: from + packet ID identifies a message. Repeats may still be acknowledged, but they are not shown twice in chat.

Direct messages use X25519 and authenticated AES-CCM instead of the channel key. Both nodes must first learn each other’s public key from NodeInfo.

Region and channel

  • Region slots: each of the 24 region presets selects a Meshtastic frequency slot, bandwidth, spreading factor, coding rate, and power ceiling (e.g., US LongFast at 906.875 MHz / SF11 / BW250). GhostESP and stock nodes must agree; mismatch = no RX.
  • Sync word: 0x442B — Meshtastic private LoRa sync word (not 0x12/0x34 public). GhostESP sets this on SX1262 at lora start; stock nodes ignore frames with other sync words.
  • Default-key channel: LongFast on chash 0x08. Custom PSKs derive a different chash; both ends must match.

Routing

  • Flood: relays wait for a short SNR-weighted window. Hearing another copy can cancel an unnecessary transmission.
  • Destination: a packet addressed to this node is delivered and acknowledged, not rebroadcast.
  • Reliable messages: the sender makes up to three attempts. The delay is calculated from modem airtime; on LongFast it is normally several seconds.
  • Next hop: successful ACKs can teach a preferred relay. If that route fails, the final retry falls back to normal flooding.
  • Duty guard: regional guard throttles forwarded TX when the band duty budget is exhausted.

NodeInfo

  • Interval: every 3 hours per node (plus on boot and on explicit request). GhostESP sends its long/short name, HW model, and role after lora start.
  • Discovery: first 30–60 seconds after lora start populates lora nodes as NodeInfos arrive. Nodes expire from the list only on long silence.
  • Diagnostics: lora nodes and lora diag expose last NodeInfo timestamps, SNR/RSSI, and hop counts.
  • Capacity: Heltec V3 keeps up to 200 nodes in RAM. Up to 32 important peer records are retained across reboot, prioritizing public keys and user flags.

See Commands for tuning SF/BW and Getting Started for first contact.