Recommendation ITU-R P.1812 in pure Kotlin Multiplatform: path-specific propagation prediction for point-to-area terrestrial services from 30 MHz to 6 GHz. Given a terrain profile, it predicts how much signal arrives at the far end.

val prediction = P1812.predict(
    path = TerrainPath(distancesKm, heightsM, clutterHeightsM, zones),
    frequencyGhz = 0.915,
    txHeightM = 10.0,
    rxHeightM = 1.5,
    pathCenterLatitudeDeg = 47.6,
)

println(prediction.basicTransmissionLossDb)
println(P1812.receivedPower(prediction, txPowerDbm = 30.0, txGainDbi = 2.0).value)

The International Telecommunication Union (ITU) publishes a reference implementation of P.1812 with validation data for every intermediate quantity, so each function here is checked against it directly. The fixtures are calls captured from eeveetza/Py1812 at 6c9061dd (P.1812-8) over its 19 validation profiles. Another 63 cases run the whole model from 30 MHz to 6 GHz, and the worst deviation in basic transmission loss is 3.3e-9 dB. The fixtures are compiled into commonTest, so the suite runs on every target, and scripts/ regenerates them from the pinned reference.

The model uses only kotlin.math: no expect/actual, no cinterop, and no runtime dependencies.

implementation("org.meshtastic:kp1812:<version>")

Android consumes the jvm artifact. There's no Android target, which keeps the Android Gradle Plugin and the Android SDK out of the build.

Swift takes a static Kp1812.xcframework through Swift Package Manager (SwiftPM). Each GitHub release attaches Kp1812.xcframework.zip and a Package.swift whose binaryTarget names it by URL and checksum. Copy that binaryTarget into your own package.

.binaryTarget(
    name: "Kp1812",
    url: "https://github.com/meshtastic/kp1812/releases/download/v<version>/Kp1812.xcframework.zip",
    checksum: "<checksum from that release's Package.swift>"
)

It covers iOS 15, macOS 12, and tvOS 15, device and simulator. There's no Mac Catalyst or watchOS slice: Kotlin/Native has no Catalyst target. From Swift, the API is P1812.shared.predict(...) with every argument spelled out, since Objective-C export drops Kotlin defaults. TerrainPath takes KotlinDoubleArray and KotlinIntArray.

jvm · js · wasmJs · wasmWasi · iosArm64 · iosSimulatorArm64 · iosX64 · macosArm64 · tvosArm64 · tvosSimulatorArm64 · linuxX64 · linuxArm64 · mingwX64

Implemented: free-space and line-of-sight loss, β₀, effective earth radius, smooth-earth heights and horizon geometry, Bullington and delta-Bullington diffraction, spherical-earth first-term diffraction, diffraction for p% time, troposcatter, ducting and layer reflection, location variability, and their combination into basic transmission loss and field strength.

Not implemented: the ITU digital maps for DN50/N050. They're ITU copyright and not redistributable, so Atmosphere takes ΔN and N₀ directly and defaults to the global medians.

GPL-3.0-or-later. The P.1812 algorithm is an ITU Recommendation, and kp1812 is an independent implementation of it.