Tags give the ability to mark specific points in history as being important
-
v1.1
8f2a4948 · ·## v1.1 (2023-02-21) Added code to setup Wireguard VPNs using SIP - Key material is derived dynamically as part of SIP call signaling - Exchange SDP attachments specially structured for Wireguard - Passing public key and to-be-implemented PSK derivation method - The idea of PSK is to derive it from authentication - This would leave the two ends with a unique shared key - The code works, but is still relatively simple
-
v1.0
Release: v1.0622555cf · ·## v1.0 (2022-09-18) Proper library, proper programs, HDLC assembler in/out, ... - Functional HDLC inject/extract, even from/to HDLC assembler - No actual HDLC negotiation yet - Proper library design: base, codec, hdlc, [back-inih/...] - Tooling now concentrated around this design - Manual pages for commands, API, and the project as a whole - Internet-Draft to level 01, begun on 02 - Considerations about cryptography - Nicely balanced / mirrorred operation of HDLC, unidirectional mode
-
v0.6
Release: v0.6e28a9ec3 · ·Rounding off proof of concept phase, with documented plans - The general model is to connect to services via network, devices, ... - Addresses and Dynamic Service Negotiation have been documented - Default HDLC info fields can go to 256 bytes (negotiate more if needed) - The FCS algorithm will be CRC-16/6sub8, HD(6) up to 8 and HD(4) up to 994 - Documented framework for MRU, MTU, network packet fragmentation, windowing - Services that cannot use fragmentation may fall back on streaming data - Services that cannot use windowing may send REJ instead of SREJ for order - Looked into a large number of services that could make sense
-
v0.5
Release: v0.534f5f210 · ·## v0.5 (2022-06-04) Adding support for 1-bit spaces for exponent 0x40 - Shift for 0x40 is << 4 _but_ for 0x00 it is << 1 - So... the lowest bit is not encoded and so we can go up to the fifth - There are interactions between pausing (zeroing the LSB) and mangling - Fixed a bug or two
-
v0.4
Release: v0.4b27c8887 · ·## v0.4 (2022-06-03) Taking care of mangled codes due to A-μ-A translation - Such bits carry no data in the lowest bit - Switched A-law data bits writing from MSB to LSB direction - This is also applied in pauses without data to transmit - Added tool/g711mangle.c to mangle as under A-μ-Α translation - Transmitted files get received with or without this mangling - Added some amateur poetry in sound/*.txt as test payload - Wrote some text about building and playing in README.md - Suggested md5sum for comparing transmitted/received files
-
v0.3
Release: v0.339c7bad2 · ·## v0.3 (2022-05-31) Reception of HDLC frames from the A-law codec - A number of files can be extracted, as though they were HDLC frames - Pausing has been simplified by syncing BREAK with the low bit - Proper handling of BREAK and FLAG to start and stop HDLC frames - Note that nothing has been done yet on HDLC address, command, checksum - The data layer is now ready for interaction with the network layer
-
v0.2
Release: v0.29c87c1d6 · ·## v0.2 (2022-05-30) Transmission of HDLC frames over the A-law codec - A number of files are transmitted, as though they were HDLC frames - Proper bit stuffing, frames surrounded with FLAG - Start new transmissions with BREAK-FLAG to clear the line - After the last frame, send FLAG-BREAK to stop expectations of another frame - Reuse of FLAG without BREAK through rejuvenation when more data exists - When not transmitting, insert "0" in the lowest bit to avoid FLAG in audio - Insertion of low "0" allows more optimal pause detection at the receiver - Receiver code is not working yet, but output from insertion looks good
-
v0.1
Release: v0.13d070b98 · ·## v0.1 (2022-05-29) Demonstrations of sound modifications to A-law and μ-law transmissions, as commonly used in ISDN, the telephony backbone, and the de-factor codec in VoIP phones. - Made a recording of The Quick Brown Fox... at normal and loud voice levels - Inserted random bits in the bits destined to carry data, causing noise - The noise level sounds like an old analog line, bearable at both levels - The noise level is clearly audible, certainly at the normal voice level - Inserted zero in the lowest bits to pause data without flag detection - The influence of these lowest bits is not audible at both levels - The data rate at the loud voice level is 27% of 64 kb/s = 17.280 kb/s - The data rate at the normal voice level is 45% of 64 kb/s = 28.800 kb/s - The data rate at complete silence level is 50% of 64 kb/s = 32.000 kb/s - Note that calls usually consist of half-time speaking (per direction) - Note that listeners are not used to silence; codecs send "comfort noise" - The experiments give rise to the work on actually inserting data frames