PaperlayerView exact revision on arXiv

Source-linked AI summary

The LQ Digital Mode Family: Protocol Architecture and Reference Specification for Weak-Signal Communications

Luis Quesada

arXiv:2609.06427v3cs.NIcs.ITeess.SPmath.IT

TL;DR

Conventional weak-signal exchanges cannot fit both a grid locator and signal report into a 77-bit reply when using fixed-width message headers. LQ8 uses variable-length prefix codes to pack them together; its four-slot protocol preserves the full bidirectional exchange and supports simultaneous pileup confirmations.

  • Problem

    A 77-bit payload can hold the two callsigns, a 15-bit locator, and a 5-bit report in 76 information bits, leaving too little space for conventional 3- or 4-bit headers.

    Source: A. Design Motivation & Prefix-Free… · p. 3

  • Method

    LQ8 uses variable-length prefix codes matched to physical-layer bit budgets to fit a locator and measured signal report into one reply frame.

    Source: §I · p. 1

  • Results

    Four time slots (60 seconds versus 5–6 messages / 75–90 seconds) provide bidirectional callsigns, locators, and reports with mutual verification; continuous pileups reach 240 QSOs/h with 0.0 dB power penalty.

    Source: A. Standard Four-Message Exchange · p. 2, C. DXpedition Fox and Hound… · p. 2

  • Takeaways & Limitations

    The protocol family also defines LQ16, LQ4, and LQ2 profiles tailored to ultra-weak channels, rapid contesting, and fast-burst links.

    Source: §I · p. 1

Abstract

from arXiv · show

This paper introduces LQ8, an open-source weak-signal digital mode for amateur radio that completes structured two-way contacts in 4 transmissions (60 s), achieving a 1.50x speedup over canonical 6-transmission exchanges (90 s) and 1.25x over 5-transmission RR73 exchanges (75 s) while preserving complete bidirectional exchange of Maidenhead grid locators, signal reports, and mutual acknowledgments. By replacing fixed-width message-type headers with variable-length prefix codes matched to physical-layer bit budgets, LQ8 packs the answering station's grid locator and measured signal report into a single 77-bit reply. In multi-station pileup operations, LQ8 confirms up to two answering stations simultaneously within a single-carrier 50.0 Hz transmission with a 0.0 dB RF power-splitting penalty, achieving a peak rate of 240 QSOs/h in continuous pileup (4x over single-carrier FT8; 480 QSOs/h dual-carrier) while retaining -21.0 dB SNR sensitivity. The protocol family also defines three extended profiles (LQ16, LQ4, and LQ2) tailored for extreme sensitivity, VHF/UHF contesting, and fast-burst channels.

I. INTRODUCTION

LQ8 addresses the 77-bit payload constraint behind conventional weak-signal exchanges by using variable-length prefix codes to fit a locator and signal report into one reply. The paper presents an architecture and software suite alongside three complementary profiles.

I. INTRODUCTION

  • Four transmissions (60 seconds) replace the conventional six-transmission (90 seconds) or shortened five-transmission (75 seconds) exchange while preserving full bidirectional contact information and mutual verification.The design retains continuous-phase 8-GFSK and LDPC(174,91) with −21.0 dB SNR sensitivity.Source: §I · p. 1

A. Architectural Philosophy & Design Principles

  • A 1-bit prefix for the dominant reply format leaves 76 bits for payload, fitting the answering station’s grid locator and measured signal report in one frame.The architecture retains clear-text callsigns across 1-on-1 exchanges and reserves compact hashes for target addressing and multi-station frames.Source: §I · p. 1, A. Architectural Philosophy & Design… · p. 1

B. Implementations & Reference Software

  • The open-source C++ reference library includes an automated verification harness and golden-vector test suites across all four modulation profiles.The paper also describes qFT8 for Android with LQ8 and live testbeds for LQ16, LQ4, and LQ2.Source: B. Implementations & Reference Software · p. 1

II. QSO PROTOCOL

The LQ protocol defines a four-message exchange that completes a contact only after both stations exchange reports and acknowledgments. It also specifies repeat behavior, logging points, and a timeout for unacknowledged attempts.

A. Standard Four-Message Exchange

B. Retransmission, Error Recovery and Logging Semantics

  • If a report or final 73 is missed, the stations repeat CALL or REPORT+73 as appropriate; a watchdog aborts after three unacknowledged attempts.The repeat exchange avoids duplicate logging when Station B reacknowledges a resent report.Source: B. Retransmission, Error Recovery and… · p. 2

C. DXpedition Fox and Hound Multi-Station Pileup Operation

LQ's prefix-coded message architecture fits richer payloads into the 77-bit frame, while Fox-and-Hound operation confirms two stations per transmission and queues callers for continued pileup service.

C. DXpedition Fox and Hound Multi-Station Pileup Operation

  • Fox confirms two Hounds in one MULTI-REPORT+73 transmission after simultaneous split-frequency calls, and both Hounds then send concluding 73 acknowledgments.The four slots are directed CQ, simultaneous Hound calls, Fox confirmation, and Hound acknowledgments.Source: C. DXpedition Fox and Hound… · p. 2
  • 240 QSOs/h in continuous pileups (4× over single-carrier FT8) is the peak rate, with 0.0 dB power penalty; dual-carrier reaches 480 QSOs/h.The passage also reports 160 QSOs/h in isolated three-slot exchanges and contrasts LQ with multi-carrier FT8's −3.0 dB RF loss.Source: C. DXpedition Fox and Hound… · p. 2

D. Multi-Caller Contention and Pileup Queuing

  • In 1-on-1 contention, Station A selects one caller while other callers retry in the next alternating slot; 24-bit callsign hashes and canonical precedence disambiguate frames.For single-target Type 11, the second target duplicates the first and decoders process it as a single-station transmission.Source: D. Multi-Caller Contention and Pileup… · p. 2, C. DXpedition Fox and Hound… · p. 2
  • A 16-station opportunistic queue uses priority aging to balance throughput and fairness, then directs REPORT+73 to the highest-priority caller after a decoded 73.A full queue evicts lowest-SNR entries below decode threshold; entries expire after four unacknowledged cycles.Source: D. Multi-Caller Contention and Pileup… · p. 2

III. MESSAGE ARCHITECTURE

  • LQ replaces fixed-width fields with entropy-matched variable-length message coding to use the 77-bit frame efficiently.Conventional headers leave only one bit beyond the 76 information bits needed for two callsigns, a grid locator, and an SNR report; uniform 3- or 4-bit headers exceed that margin.Source: §III · p. 3, A. Design Motivation & Prefix-Free… · p. 3

A. Design Motivation & Prefix-Free Framing

  • A one-bit prefix for the dominant CALL STD (NOSUF) format preserves 76 payload bits, combining station identification, locator delivery, and signal-report measurement in one frame.Simpler messages such as standard CQ require 43 payload bits and can accommodate longer prefix codes.Source: A. Design Motivation & Prefix-Free… · p. 3
  • The receiver identifies message types by traversing a binary prefix tree bit by bit, while Types 14–16 remain reserved for backward-compatible expansion.The active operational codebook is prefix-free but intentionally not Kraft-complete.Source: A. Design Motivation & Prefix-Free… · p. 3

B. Operational Categories & Message Catalogue

The catalogue defines operational message formats and bit-level conventions, while boundary rules specify canonical format selection, disambiguation, and compatibility behavior.

C. Protocol Boundary Conditions & Canonical Precedence

  • Canonical precedence selects the most compact valid format, with Type 5 reserved for standard callsigns without suffixes and other types handling non-standard or portable callsigns.Type 6 carries the caller’s locator and SNR; Type 7 sends a non-standard caller’s clear-text identity to a target hash.Source: C. Protocol Boundary Conditions &… · p. 3
  • In Fox mode, Type 11 or 12 duplicates Target 1 when confirming one station, while Type 8 supersedes Type 11 in one-on-one contacts.Single-target Type 11 is reserved for non-standard callers or split Fox-and-Hound schedules.Source: C. Protocol Boundary Conditions &… · pp. 3–4
  • Uniform 24-bit callsign hashes yield a collision probability of at most 0.75% across 500 stations, with local caches and canonical precedence used for disambiguation.In one-on-one contacts, standard stations ignore hashed completions from standard peers.Source: C. Protocol Boundary Conditions &… · p. 3
  • Compound callsigns up to 13 characters use Type 4 CQ frames or free text for directed replies; the reference suite verifies all 13 formats across channel conditions.Decoders treat unassigned Types 14–16 as framing boundaries and safely ignore them.Source: C. Protocol Boundary Conditions &… · pp. 3–4

B. Operational Categories & Message Catalogue

IV. FIELD ENCODINGS

LQ adopts core field structures from FT8 while specifying a canonical 28-bit representation for standard callsigns and a mixed-radix encoding.

IV. FIELD ENCODINGS

  • LQ adopts FT8’s 28-bit callsigns, 15-bit grid locators, 20-bit CQ modifiers, and 5-bit signal reports, while adding fields for portable suffixes and non-standard callsigns.The paper also introduces CRC-24/Q hashes and a 104-character free-text Varicode.Source: §IV · p. 5

A. Standard Callsign (28 bits)

  • Standard ITU callsigns are canonicalized as six characters around a decimal digit at position three; callsigns with slashes or other formats are non-standard.The canonical pattern is c1c2ds1s2s3, with the prefix right-aligned and suffix left-aligned using spaces.Source: A. Standard Callsign (28 bits) · p. 5
  • A mixed-radix mapping packs the six canonical character indices into a 28-bit integer N28.The character positions use radices 37, 36, 10, 27, 27, and 27.Source: A. Standard Callsign (28 bits) · p. 5
  • Valid standard callsigns occupy N28 values through 262,177,559, followed by reserved values for DE, QRZ, and CQ; decoders validate radix ranges.The reserved values are 262,177,560, 262,177,561, and 262,177,562, respectively.Source: A. Standard Callsign (28 bits) · p. 5

B. Standard Dedicated Suffixes (1 bit)

The field encodings add a dedicated portable suffix, Base-38 packing for non-standard callsigns, and hash-based addressing for directed and multi-station frames.

B. Standard Dedicated Suffixes (1 bit)

C. Non-Standard Callsigns (Base-38 Big-Integer Packing)

  • Base-38 packing supports non-standard callsigns up to nine characters in 48 bits, or compound callsigns up to 13 characters in 69 bits.The 13-character form is used in CQ frames; the alphabet includes letters, slash, digits, and space.Source: C. Non-Standard Callsigns (Base-38… · p. 5

D. Callsign Hashes (24-bit, 20-bit, and 16-bit) & Dehashing

  • CRC-24/Q produces 24-bit callsign hashes, while directed Type 7 replies use the 20 most significant bits; multi-station Types 11 and 12 use 16-bit DX hashes.For 16-bit dehashing, decoders restrict resolution to actively called stations, evicted upon QSO completion or timeout.Source: D. Callsign Hashes (24-bit, 20-bit,… · p. 5

E. Maidenhead Grid Locator (15 bits)

The field encodings pack Maidenhead locators into 15 bits and use 20-bit modifiers to distinguish CQ types and designators.

E. Maidenhead Grid Locator (15 bits)

F. CQ Modifiers (20 bits)

  • A 20-bit CQ modifier encodes general CQ, three-digit numeric designators, or alphanumeric tokens up to four characters.Longer modifiers use four-character abbreviations or free text.Source: F. CQ Modifiers (20 bits) · p. 5

G. Signal Reports (5 bits)

The section specifies how LQ8 represents measured signal reports and free text, and describes the physical-layer and channel-coding components it retains.

G. Signal Reports (5 bits)

  • Reports encode measured SNR from −26 dB to +5 dB as Nsnr = clamp(SNRdB + 26,0,31), with R prefixed in acknowledgments.Directed CALL replies display the report without a prefix.Source: G. Signal Reports (5 bits) · p. 5

H. Free-Text Varicode Encoding (73 bits)

  • Type 13 free text averages 4.25 bits per character and ≈17.2 characters; Figure 1 describes an 8-GFSK, 50.0 Hz, 12.64 s frame.The 104-character prefix code spans uppercase letters, digits, punctuation, Latin-1 characters, newline, and fill; extraction ends at the first zero bit.Source: H. Free-Text Varicode Encoding (73 bits) · p. 5, Figure 1 · p. 6

V. PHYSICAL LAYER

  • LQ8 adopts FT8’s continuous-phase modulation and error-correction coding to retain −21.0 dB SNR sensitivity in a 2500 Hz bandwidth.Source: §V · p. 6

A. Channel Coding: CRC-14 and Systematic LDPC(174,91)

  • The channel-coding chain appends a 14-bit CRC to the 77-bit payload, pads the block, then uses systematic LDPC(174,91) to add 83 parity bits.The CRC parameters are specified in Table 5; the Gallager generator matrix is described in Table 6.Source: A. Channel Coding: CRC-14 and… · p. 6, Table 5 · p. 6

B. LQ8 Physical Transport & UTC Slot Framing

LQ8 uses FT8-derived 8-GFSK transport in 15-second UTC slots, with continuous-phase synthesis and defined tone mapping.

B. LQ8 Physical Transport & UTC Slot Framing

C. Modulation Mapping & Continuous-Phase Synthesis

  • A 3-bit Gray code maps bit triplets to eight tones, and continuous-phase audio synthesis uses half-symbol raised-cosine envelope shaping and Gaussian frequency trajectories.The LQ8 ramp is 80.0 ms; the frequency pulse is truncated to 3Ts.Source: C. Modulation Mapping &… · p. 6

D. Costas Synchronization & DSP Pipeline

LQ8 combines Costas-array synchronization with a multi-pass receiver pipeline, while the protocol family extends to profiles for different channel and rate conditions.

D. Costas Synchronization & DSP Pipeline

  • Three 7-symbol Costas arrays isolate LQ8 and LQ16 synchronization from FT8, with no more than one coincidence hit against FT8.The 174-bit LDPC codeword is split into two 87-bit blocks without bit interleaving; the receiver uses matched filtering, soft LLRs, LDPC decoding, CRC verification, and candidate subtraction.Source: D. Costas Synchronization & DSP Pipeline · p. 6
  • Table 6 specifies the Gallager LDPC generator matrix as 83 rows of 23 hexadecimal digits, with one unused padding bit per row.Source: Table 6 · p. 7

VI. EXTENDED MODES: LQ16, LQ4, AND LQ2

  • LQ16 targets ultra-weak paths at −24.0 dB, while LQ4 and LQ2 target faster contesting and burst links, with LQ2 reaching 960 QSOs/h Fox/Hound.LQ4 uses 7.5 s slots and 30.0 s contacts; LQ2 uses 3.75 s slots and 15.0 s contacts.Source: §VI · p. 7

A. 4-GFSK Frame Structure & Modulation

  • The LQ4 and LQ2 frame uses 105 continuous-phase 4-GFSK symbols, with three data blocks separated by four Costas sync sequences.The 77-bit payload is whitened before CRC-14/LDPC encoding; the 174-bit codeword maps in consecutive pairs to 4-GFSK tones, with guard intervals for propagation and turnaround.Source: A. 4-GFSK Frame Structure & Modulation · p. 7

B. Profile Selection and Channel Matching

The section matches operational profiles to channel conditions and reports faster contacts or greater sensitivity across the LQ family. It also describes the suite’s implementation and bounds peak-rate claims by their operating assumptions.

B. Profile Selection and Channel Matching

VII. CONCLUSION

  • LQ8 completes confirmed contacts in 4 transmission slots (60 s), retains −21.0 dB SNR sensitivity, and confirms two pileup stations at up to 240 QSOs/h single-carrier.Dual-carrier operation reaches 480 QSOs/h.Source: §VII · p. 8
  • LQ4 and LQ2 use 4-GFSK and Costas sync arrays for 30.0 s and 15.0 s contacts, supporting burst rates up to 960 QSOs/h on VHF/UHF meteor scatter and sporadic-E openings.Source: §VII · p. 8
  • The open-source C++ reference library includes an automated verification harness with > 90% line coverage and golden-vector tests; qFT8 field verification spans HF portable activations.Source: §VII · p. 8
Loading 2609.06427v3…