Packet Radio

AX.25: The Data Link Layer

AX.25 (Amateur X.25) is a data-link-layer protocol adapted from ITU-T X.25 for amateur radio by TAPR (Tucson Amateur Packet Radio) in 1984. It provides connectionless and connection-oriented data transfer between amateur stations using HDLC-like framing over a physical radio channel. Unlike wired networks, the shared RF medium means every station hears every transmission, requiring careful collision management.

Frame Structure

Every AX.25 frame is delimited by 0x7E flag bytes. The frame is followed by an address field (destination + source + up to eight digipeaters), a control field, an optional information field, and a 16-bit CRC-16 (FCS) for error detection. The maximum information field length is 256 bytes, yielding an overall frame size typically under 330 bytes at the physical layer.

AX.25 borrows HDLC's bit-stuffing transparency mechanism: whenever five consecutive 1-bits appear in the data, a 0-bit is automatically inserted. This prevents data from being misinterpreted as the 0x7Eflag. On reception, stuffed bits are stripped. The 0x7E flags may overlap between consecutive frames, eliminating inter-frame gaps and allowing continuous synchronous transmission.

Address Field Encoding

The AX.25 address field is structured for amateur radio callsigns. Each callsign occupies 7 bytes: the 6-character callsign (padded with spaces if shorter) is shifted left by one bit and stored in the first 6 bytes, while the 7th byte contains the SSID (Sub-Station ID, 0–15) in bits 1–4, a reserved bit, a command/response bit, and an extension bit (bit 7) that is set on the last address entry. The address field begins with the 7-byte destination, followed by the 7-byte source, and zero or more 7-byte digipeater addresses.

Callsigns are encoded as ASCII shifted left by one bit. For example,WB2ABC becomes 0x56 0xC5 0x62 0xC1 0xC2 0xC3in the address bytes. The trailing SSID byte includes the SSID value, command/response flag, and the H-bit (has-been-repeated) used by digipeaters to suppress duplicate forwarding.

Control Field

The control field identifies the frame type. It is a single byte that distinguishes between I (Information), S (Supervisory), and U (Unnumbered) frames. The two least-significant bits of the control byte determine the frame class:

• I-frames (bit 0 = 0): Carry sequenced data. The control byte contains N(S) (send sequence number, bits 1–3 and 6), N(R) (receive sequence number, bits 5–8), and a poll/final (P/F) bit (bit 4). I-frames support flow control and error recovery.
• S-frames (bits 0–1 = 01): Supervisory frames for acknowledgements and flow control. Types: RR (Receive Ready), RNR (Receive Not Ready), REJ (Reject), SREJ (Selective Reject). Each carries an N(R) and P/F bit.
• U-frames (bits 0–1 = 11): Unnumbered frames for connection management. No sequence numbers. Includes SABM (Set Asynchronous Balanced Mode), UA (Unnumbered Acknowledge), DM (Disconnected Mode), DISC (Disconnect), UI (Unnumbered Information), and XID (Exchange Identification).

Connection Modes

AX.25 supports two fundamental modes. Connected modeestablishes a reliable virtual circuit using SABM/UA handshake. The two stations exchange I-frames with sequence numbers, and the receiver acknowledges with RR or REJ frames. If a frame is lost, the sender retransmits. DISC/UA tears down the connection. Unconnected mode uses UI frames for one-way broadcasts with no delivery guarantee—ideal for APRS and beacon transmissions.

KISS TNC: The Computer Interface

A Terminal Node Controller (TNC) is the modem that converts digital data into audio tones (AFSK) suitable for transmission over FM or SSB radio. The KISS protocol, defined in TAPR's RFC, provides a simple byte-oriented interface between a computer application and the TNC. It abstracts the modem layer, allowing software to send and receive raw AX.25 frames.

KISS frames are delimited by FEND (0xC0). The command byte follows the opening FEND: its low nibble selects the TNC port (allowing multi-port TNCs) and the high nibble selects the command (0 = data, 1 = TX delay, 2 = TX tail, 3 = full duplex, etc.). The frame payload is simply the raw AX.25 frame bytes.

Byte transparency is handled by FESC (0xDB). Whenever aFEND or FESC byte appears in the data, it is replaced: FEND → FESC TFEND (0xDB 0xDC) andFESC → FESC TFESC (0xDB 0xDD). On reception, the receiver reverses this substitution.

Frequency Bands for Packet Radio

VHF — 2 Meters (144–148 MHz)

The most active packet radio band worldwide. In the United States,144.390 MHz is the designated APRS frequency, carrying thousands of stations simultaneously. Most packet activity on 2m uses 1200 baud Bell 202 AFSK (Audio Frequency Shift Keying) with ±100 Hz deviation for FM simplex. The 2m band offers line-of-sight propagation with significant ground-wave range—typically 10–50 miles depending on terrain, antenna height, and power. Digipeaters on hilltops extend this range to hundreds of miles for APRS traffic.

In other regions, the APRS frequency varies: 144.390 MHz in North America, 144.800 MHz in Europe (CEPT), 144.640 MHz in Japan, and 144.350 MHz in Australia. Packet radio on 2m also supports 9600 baud (G3RUH compatible) using direct FSK on FM transceivers, though this requires radio modifications and is less common.

UHF — 70 Centimeters (420–450 MHz)

The 70cm band hosts packet activity at 430.500 MHz in regions where APRS operates on UHF. In North America, 70cm packet is less common on APRS but is used for point-to-point links, mesh networking, and emergency communications. Higher frequency means shorter wavelength, which allows smaller directional antennas—advantageous for portable packet stations. Some operators run 9600 baud packet on 70cm for higher throughput than the 1200 baud 2m standard.

HF (3–30 MHz)

HF packet radio is fundamentally different from VHF/UHF: it operates over sky-wave propagation, bouncing signals off the ionosphere for worldwide reach at low power. Common HF packet frequencies include7.035 MHz (40m, LSB), 10.150 MHz (30m),14.070 MHz (20m), 18.100 MHz (17m),21.100 MHz (15m), and 28.120 MHz (10m). HF propagation is unreliable—signals fade in and out with ionospheric conditions, so HF packet requires robust protocols with forward error correction. Standard AX.25 is marginal on HF; protocols like PACTOR, WINMOR, and VARA use convolutional coding and ARQ to maintain throughput.

Packet Formats

AX.25 defines three fundamental frame types that mirror the OSI data-link layer classification. Understanding these is essential for working with packet radio software and diagnosing problems with packet networks.

UI Frames (Unnumbered Information)

UI frames carry data without establishing a connection. They are the workhorse of APRS and beacon systems. A UI frame has the control byte set to0x03 (P/F = 0) or 0x13 (P/F = 1). Because there is no connection state, any station can transmit a UI frame at any time. The information field contains the payload (up to 256 bytes). UI frames arenot acknowledged—if the receiver misses it, the data is lost. This makes UI frames ideal for position reports, weather data, and messages that tolerate occasional loss.

I Frames (Information)

I frames carry sequenced data within an established AX.25 connection. Each I-frame carries a send sequence number N(S) that increments with each successive frame, and a receive sequence number N(R) that acknowledges all frames up to N(R)−1. The window size is 8 (modulo 8 arithmetic), meaning up to 7 unacknowledged frames may be in flight. If a frame is received out of order, the receiver sends a REJ frame requesting retransmission from the missing frame onward. I-frames support flow control through RNR frames—if the receiver's buffer is full, it sends RNR to halt transmission.

S Frames (Supervisory)

S-frames carry no information field. They manage the connection state:

• RR (Receive Ready, code 00): Acknowledges received frames up to N(R)−1 and indicates readiness to receive more data.
• RNR (Receive Not Ready, code 01): Acknowledges frames but temporarily halts further I-frame transmission. The sender must poll with RR(P=1) to resume.
• REJ (Reject, code 10): Requests retransmission starting at frame N(R). Used in go-back-N ARQ.
• SREJ (Selective Reject, code 11): Requests retransmission of a specific frame without discarding correctly received frames. Not all implementations support SREJ.

U Frames (Unnumbered)

U frames control the connection state machine. The modifier bits in the control byte (bits 2, 3, 5, 6, 7) define the U-frame subtype:

• SABM (Set Asynchronous Balanced Mode): Initiates a connection. The sender transmits SABM(P=1); the receiver responds with UA(F=1) to accept or DM(F=1) to reject.
• UA (Unnumbered Acknowledge): Confirms SABM or DISC.
• DM (Disconnected Mode): Rejects SABM or indicates the station is not connected.
• DISC (Disconnect): Terminates an established connection. The peer responds with UA.
• XID (Exchange Identification): Negotiates parameters between stations.

APRS: Automatic Packet Reporting System

APRS (Automatic Packet Reporting System) is a real-time digital communications system built on top of AX.25 UI frames. Conceived by Bob Bruninga (WB4APR) in 1992, APRS transforms packet radio from a point-to-point messaging system into a global information network. APRS packets are not acknowledged—instead, they are digipeatedacross the network, creating a broadcast topology where all stations receive all traffic within RF range.

Callsign-SSID Addressing

APRS uses the standard AX.25 callsign-SSID addressing system. A station is identified by its callsign and an optional SSID (0–15). The SSID allows a single operator to run multiple virtual stations—one for a vehicle, one for a weather station, one for a home base, etc. Conventions: SSID 0 is the primary station, SSID 7 is for HF APRS, SSID 10–14 are for mobile/portable stations, and SSID 15 is experimental. APRS addresses also include a tocall (destination address) that identifies the software or device generating the packet—for example, APJWC1 for APRSISCE/32 or APW32Nfor Winlink.

Digipeater Paths

APRS packets include a digipeater path in the AX.25 address field. The path specifies a chain of digipeaters that will relay the packet. Standard paths use WIDE1-1,WIDE2-1 (New-N paradigm) to allow fill-in digipeaters and wide-area digipeaters to collaborate. Each digipeater that relays the packet sets the H-bit in its address to prevent duplicate forwarding. The New-N paradigm replaced the older RELAY,WIDE path in the early 2000s to reduce network congestion and improve reliability.

A digipeater hears a packet, checks the path, and if its callsign appears in the next position, it decrements the hop count and retransmits. If the path is WIDE2-1, the digipeater changes it to WIDE2-0 before forwarding— after which no further WIDE2 digipeaters will relay it. WIDE1-1 is replaced with WIDE2-0, allowing fill-in digipeaters to participate without extending the path indefinitely.

Packet Types and Data Formats

APRS defines a rich set of data types identified by a single character data type identifier in the information field. Common types include:

• Position (! / = / @ / /): Latitude/longitude with optional altitude, course, speed, and station comments. Uses compressed or uncompressed APRS coordinate format.
• Message (:): Direct text messages between stations. Messages include a addressee (9 characters, space-padded), the message text, and a message ID for acknowledgement. The sending station retries up to 5 times; the receiving station sends an ack (:addressee:ackNNN).
• Object (;) and Item ()): Named items with positions, used for events, hazards, weather stations, and search-and-rescue targets.
• Weather (_) and Mic-E (_): Weather reports from automated stations and Mic-E-encoded position/weather data from Kenwood handhelds.
• Telemetry (T): Analog and digital sensor data with calibration parameters.
• Status (> / {): Free-form status text, often from emergency operations centres.

AIS Integration

APRS has been integrated with Automatic Identification System (AIS) data from marine vessels. AIS receivers decode VHF marine band AIS packets (161.975 and 162.025 MHz) containing ship positions, course, speed, and identification. This data can be reformatted as APRS packets and injected into the APRS-IS internet network, allowing marine vessel tracking on APRS maps. Several coastal APRS stations run dual-mode receivers that decode both APRS and AIS, creating a unified real-time picture of both amateur radio stations and maritime traffic. This integration is particularly valuable for coastal emergency operations, port security, and marine search and rescue coordination.

APRS-IS: The Internet Backbone

While APRS is fundamentally a radio-based system, the APRS Internet Service (APRS-IS) connects radio-based APRS to the global internet. IGates (Internet Gates) receive APRS packets over RF and forward them to regional APRS-IS servers, where they are aggregated and made available via the APRS-IS database. APRS clients and web-based mapping tools query this database to display real-time positions worldwide. IGates can also inject internet-received APRS messages onto RF, enabling internet-to-radio message delivery for stations without direct RF contact.

BBS and Mail: Store-and-Forward Networking

Packet radio bulletin board systems (BBS) are the backbone of store-and-forward messaging in amateur radio. A BBS station receives messages over RF, stores them on disk, and retransmits them on request or automatically to other BBS nodes. This architecture enables email-like communication over radio networks without internet connectivity.

BPQNode and FlexNet

BPQNode (originally by John Wiseman, G8BPQ, now maintained as BPQ32) is a software framework that implements AX.25 packet switching on PC platforms. It provides node-to-node routing, BBS services, and gateway functions between packet radio and the internet. BPQNode supports NET/ROM and FlexNet routing protocols that automatically discover and maintain routes through a mesh of packet radio nodes. A BPQNode system typically connects to a TNC via serial port or KISS interface, and serves as both a user-accessible node (via AX.25 connected-mode sessions) and a network router forwarding traffic between nodes.

Winlink

Winlink (formally Winlink 2000) is the world's largest amateur radio email system, operating since 1986. It provides full-featured email with attachments, using HF, VHF, and UHF packet radio as the transport. Winlink operates a network of CMS (Common Message Servers) accessible via internet or radio. Users compose email in a client application (such as Pat, Winlink Express, or Airmail), which connects to a local Winlink gateway station (a BBS connected to the internet) or directly to a CMS via HF radio. The gateway or CMS delivers the email to the internet email system, and vice versa.

Winlink uses proprietary transport protocols over AX.25: PACTOR(I–IV), WINMOR, and VARA for HF, and standard AX.25 packet for VHF/UHF. These protocols use convolutional coding, ARQ, and adaptive modulation to maintain throughput over unreliable HF links. Winlink is widely used for emergency communications (EMCOMM) when internet infrastructure is unavailable—during hurricanes, wildfires, and other disasters, Winlink operators provide email capability for emergency management agencies.

Winlink traffic statistics are impressive: the network processes over 10 million messages per month globally. Each message is routed through a network of gateway stations and CMS servers, with automatic store-and-forward ensuring delivery even when direct paths are unavailable. The system also supports tactical addressing—stations can use any valid callsign with a tactical address (like NE-BEACH-OPS) for operational flexibility during emergencies.

Modern Digital Modes over Packet

The evolution of digital modes on amateur radio has been driven by the need to push more data through increasingly narrow and noisy channels. While traditional AX.25 packet radio at 1200 baud Bell 202 AFSK remains the backbone of VHF/UHF packet, HF operation demands more sophisticated modulation and coding schemes.

Winlink Transport Protocols

Winlink has developed several transport protocols optimised for different channel conditions. PACTOR (developed by SCS in Germany) uses a combination of PSK (Phase Shift Keying) and FSK with convolutional coding and ARQ. PACTOR-I achieves about 100–200 bps on HF; PACTOR-IV reaches 500–2000 bps with adaptive modulation.WINMOR is an open-source protocol using 4-FSK modulation with rate 1/2 convolutional coding and ARQ, achieving 200–1600 bps depending on conditions.VARA (developed by José Alberto, EA5HJZ) uses OFDM (Orthogonal Frequency Division Multiplexing) with LDPC (Low-Density Parity-Check) coding, achieving 500–3500 bps on 2.5 kHz SSB bandwidth— among the highest throughput of any HF digital mode. VARA is now the preferred Winlink transport for most HF operations.

JS8Call: Weak-Signal Keyboard Messaging

JS8Call, developed by Jordan Sherer (KN4ITG), is a keyboard-to-keyboard messaging mode derived from FT8 (the contest and DX mode by K1JT). JS8Call uses 8-FSK modulation with a 20-second transmission cycle, achieving reliable copy at signal-to-noise ratios as low as −24 dB. This means JS8Call can decode signals buried 24 dB below the noise floor—invisible to the human ear. JS8Call supports directed callsigns, directed messaging, and automatic relay nodes. It operates on HF bands (typically 40m, 30m, 20m, 17m, 15m) and has become popular for emergency communications and low-power (QRP) operation where signals are marginal.

VARA: High-Speed HF Digital

VARA represents the current state of the art for HF digital communication. Unlike traditional narrowband modes, VARA uses OFDM with 32 subcarriers spaced across 2.5 kHz bandwidth, with each subcarrier modulated using 4-PSK or 8-PSK depending on channel conditions. Forward error correction uses LDPC codes at rates from 1/2 to 3/4. The combination of adaptive modulation, powerful FEC, and ARQ allows VARA to sustain 2–3.5 kbps on a single SSB channel. VARA is available in both Winlink-proprietary (for Winlink email) and open-source (for general packet radio) versions.

FT8 and Packet Integration

FT8 (by K1JT) has revolutionised weak-signal work, but its fixed 15-second transmission cycle and 75-character message limit make it unsuitable for general packet radio. However, FT8's modulation and coding techniques have influenced modern packet radio protocols. Some systems integrate FT8 as a beacon mode for weak-signal position reporting on HF. WSJT-X software by K1JT implements FT8, FT4, MSK144, and other weak-signal modes; while these are distinct from packet radio, they occupy adjacent frequency segments on HF and serve complementary roles in amateur digital communications.

Packet Radio Network Topology

Packet radio networks are organised around three fundamental node types. Understanding these is essential for designing and operating packet infrastructure.

• Digipeaters: RF-to-RF relay stations that receive a packet and retransmit it on the same frequency. Digipeaters are typically placed on hilltops or tall buildings for maximum coverage. They set the H-bit in the digipeater address to prevent duplicate relaying. APRS digipeaters operate on VHF and UHF; HF digipeaters extend coverage over continental distances.
• Nodes: Software systems (BPQNode, FlexNet, NET/ROM) that provide multi-port routing and user access. A node typically connects multiple radio ports and an optional internet gateway. Users can connect to a node via AX.25, send messages, and the node routes traffic to other nodes automatically. Node networks can span continents— operators have maintained packet networks stretching from coast to coast in the United States using chains of node stations.
• Mailboxes (BBS): Store-and-forward systems that hold messages for later retrieval. A BBS receives messages over RF, stores them, and serves them to users who connect directly. BBS systems also forward messages to other BBS stations, creating a store-and-forward email network independent of the internet. Winlink CMS servers are the modern evolution of this concept—centralised, high-capacity message servers accessible via both radio and internet.

Emergency Communications

Packet radio's greatest value may be its role in emergency communications. Unlike cellular and internet infrastructure, packet radio networks operate independently of power grids and commercial telecommunications. APRS provides real-time situational awareness—position tracking, weather data, and status reports from field operators. Winlink provides email capability when internet is unavailable. And Winlink's HF operation allows email across continental distances with no infrastructure beyond a radio, a laptop, and a wire antenna.

During Hurricane Katrina (2005), the Japan earthquake and tsunami (2011), and numerous wildfire seasons, amateur radio operators deployed packet radio networks to provide emergency communications when all commercial infrastructure failed. APRS tracked rescue teams in real-time; Winlink delivered situation reports and resource requests. These capabilities are not theoretical—they have been proven repeatedly in real emergencies, saving lives and enabling coordinated disaster response.

The resilience of packet radio comes from its distributed architecture. There is no single point of failure. Any two stations with compatible equipment can establish a data link. Digipeaters can be deployed rapidly—portable digipeaters on battery power can be set up in minutes. The internet can be completely destroyed, and packet radio will continue to function. This is not redundancy—it is independence.