Audio used alongside WiFi to solve contention
40% gain in throughput; 27% for dense networks
Limitations of audio - frequency, penetration
Software based

1) Contention resolution over acoustic channel
2) Data transmission over wifi

Random Tone is generated - lowest tone first, while also listen
Contention resolution
Wait for DIFS
Transmit / Wait
Then repeat simultaneously

Losers subtract their RNG number by the winner

Shared flag to lock and unlock

Single Acoustic Domain

16 - 21 kHz


Majority of background noises go up to 12 KHz
Smart devices have high acoustic sensitivity, can capture up to 21 KHz

Propagation Delay

Takes a while to decide who wins.
Allow multiple packets to be sent

Also decide the order of the next few senders

Round-robin - parallel

  • WAIT FOR DIFS -> Waits for given period of silence


Same acoustic tone, tone number

  • double round

  • length

  • tone

  • Mechanical artifacting from sudden speaker movement

Amplitude fade-in and fade-out, 5 milliseconds

Increase frequency set
gaps between frequency tone - 200 Hz gaps
26 frequencies

double round if collision occurs!! ————– 26 devices isn’t that much though

Kernel space

A-PHY - DMA; generateTone()
A-MAC - Get tone number, run Harmony, get rank, DIFS

20 meter Harmony pax

Hidden nodes