Traffic breakdown all monitored window vs last hour
Type
Packets
% all
% last 1h
Receptions (RF)
% RF
share
“Receptions” = how many observer nodes heard the packet — a proxy for airtime /
mesh load (a 20-hop flood is heard ~40×, a direct message once).
⚠️ Offenders click a row for details & an evidence pack
Sev
Kind
Who
Detail
Rate
Last
Sev
Kind
Who
Detail
Rate
Last
Sev
Kind
Who
Detail
Rate
Last
Three detection windows run simultaneously, each with its own rates. Events are raised from the 24h window.
Click column headers to sort.
🔮 What-if if an offender went quiet — mesh airtime freed and repeater duty relieved, last 24h
Sev
Kind
Who
Freed s/h
% of mesh
Relieved repeaters (peak-hour duty, % of the limit)
🔁 Repeat offenders persistent ledger — days active, from the history database
Sev
Kind
Who
Days
Snapshots
First seen
Last seen
Peak rate
🏳 Whitelist known-good sources — hidden from counts, events and alerts; stored in meshguard_whitelist.json
Subject
Scope
Note
Added
Open an offender row and use “Whitelist” to add one. Matching uses the public key or the traffic-source key, not just the display name.
🛰 GRP_DATA senders group datagrams = app beacons / telemetry / position spam
Source
Ch
Pkts
per h
App
Best candidate
Relay (first hop)
Last
Flood paths exclude the originator, so senders are inferred from relay adjacency
and advert-path neighbourhood matches. See “How attribution works” in the footer.
🎚 Calibration what "normal" looks like on this mesh right now
⏱ Time hogs who burns the most mesh airtime, last 24h
Who
Class
Pkts
s/h
% pot
% mesh
Hops
Now vs avg
vs typical
24h
Flags
Airtime by traffic type (24h, relay-amplified)
One unit for everything: LoRa time-on-air of the whole frame (header, path, payload) × (relay hops + 1), per source. “% pot” = share of one repeater’s
legal 10% duty budget (360 s/h). “Now vs avg” compares the last hour with the 24h average.
Classes: ADVERT = one node’s adverts · CONV = request/response/DM conversation between two hashes ·
GRP = a group channel via one entry relay · ANON_REQ = anonymous requests from one key.
🔋 Repeater duty cycle estimated TX airtime vs the 10% legal limit (360 s/h), last 24h
Repeater
TX 24h (s)
Peak hour (s)
% of limit
Hours over
24h
Every repeater on an observed path transmitted that packet once, so its TX time is the sum of the packets it forwarded.
This is a lower bound (observers miss some paths). A 1-byte hash shared by several nodes shows as “shared”.
📍 Entry hotspots where airtime enters the mesh (first relay), last 24h — also drawn on the Map
Entry relay
Air 24h (s)
s/h
% pot
Biggest sources
🦘 Hop count relay hops per packet
🌍 Region airtime budget 24h, by scope
Region
Air (s)
% mesh
% pot
vs 7d daily avg
📡 Relay load carried group airtime, 24h
Entry relay
Air 24h
Packets
% of pot
LoRa SF8/62.5 kHz (CR 4:8, 869.618 MHz) airtime model × relay participations. A saturated relay starves every node behind it.
Top talkers
Node
Type
Pkts
RF %
Breakdown
Seen
👀 Observer coverage who is hearing the mesh (24h)
Observer
Packets heard
Share
Heard only by this one
Group channels decoding statusclick a row to read its messages
Channel
Hash
GRP_TXT
GRP_DATA
Decoded
Status
Air 24h
Msgs
Channel keys come from: dictionary decoding (rockyou.txt, Baltic hunspell, name-days,
KiekR catalogue) · atin.lt official channel QR secrets · MeshMapper API · public-channel
constant PSK · hashtag PSK = SHA256('#name')[:16]. unknown = private random-key
channel (survived every wordlist — needs a leaked key); partial = another private
channel shares this 1-byte on-air id; undecoded = none of the candidate keys
MAC-verify this traffic — it's a private channel, and a name ending in ? is just
an unverified key-guess from the catalogue sharing that id.
💬 Channel messages read decoded traffic per channel
Time
Kind
Sender
Content
Via
Rx
Last 200 decoded messages per channel (kept in memory + state, 300 max). GRP_TXT shows
plaintext “Name: message”; GRP_DATA shows the app (data-type allocation) and payload hex.
Unknown channels can’t be read without their private key.
Region scopes flood/direct transport-code decode (HMAC vs #ee #eu #lt #lv)