AX.25 Amateur Packet-Radio
Link-Layer Protocol
Version 2.0, October 1984

2 AX.25 Link-Layer Protocol Specification

2.1 Scope and Field of Operation

This protocol conforms to ISO Recommendations 3309, 4335 (including DAD 1&2) and 6256 high-level data link control (HDLC) and uses some terminology found in these documents. It also conforms with ANSI X3.66, which describes ADCCP, balanced mode.

This protocol follows, in principle, the CCITT X.25 Recommendation, with the exception of an extended address field and the addition of the Unnumbered Information (UI) frame. It also follows the principles of CCITT Recommendation Q.921 (LAPD) in the use of multiple links, distinguished by the address field, on a single shared channel.

As defined, this protocol will work equally well in either half- or full-duplex Amateur Radio environments.

This protocol has been designed to work equally well for direct connections between two individual amateur packet-radio stations or an individual station and a multiport controller.

This protocol allows for the establishment of more than one link-layer connection per device, if the device is so capable.

This protocol does not prohibit self-connections. A self-connection is considered to be when a device establishes a link to itself using its own address for both the source and destination of the frame.

Most link-layer protocols assume that one primary (or master) device (generally called a DCE, or data circuit- terminating equipment) is connected to one or more secondary (or slave) device(s) (usually called a DTE, or data terminating equipment). This type of unbalanced operation is not practical in a shared-RF Amateur Radio environment. Instead, AX.25 assumes that both ends of the link are of the same class, thereby eliminating the two different classes of devices. The term DXE is used in this protocol specification to describe the balanced type of device found in amateur packet radio.

2.2 Frame Structure

Each field is made up of an integral number of octets (or bytes), and serves a specific function as outlined below.

2.2.1 Flag Field

2.2.2 Address Field

The encoding of the address field is described in 2.2.13.

2.2.3 Control Field

2.2.4 PID Field

The PID itself is not included as part of the octet count of the information field. The encoding of the PID is as follows:

Where:
A y indicates all combinations used.

Note:
All forms of yy11yyyy and yy00yyyy other than those listed above are reserved at this time for future level 3 protocols. The assignment of these formats is up to amateur agreement. It is recommended that the creators of level 3 protocols contact the ARRL Ad Hoc Committee on Digital Communications for suggested encodings.

2.2.5 Information Field

2.2.6 Bit Stuffing

2.2.7 Frame-Check Sequence

2.2.8 Order of Bit Transmission

2.2.9 Invalid Frames

2.2.10 Frame Abort

2.2.11 Interframe Time Fill

2.2.12 Link Channel States

2.2.13 Address-Field Encoding

The HDLC address field is extended beyond one octet by assigning the least-significant bit of each octet to be an "extension bit". The extension bit of each octet is set to zero, to indicate the next octet contains more address information, or one, to indicate this is the last octet of the HDLC address field. To make room for this extension bit, the amateur Radio call sign information is shifted one bit left.

2.2.13.1 Nonrepeater Address-Field Encoding

A1 through A14 are the fourteen octets that make up the two address subfields of the address field. The destination subaddress is seven octets long (A1 thru A7), and is sent first. This address sequence provides the receivers of frames time to check the destination address subfield to see if the frame is addressed to them while the rest of the frame is being received. The source address subfield is then sent in octets A8 through A14. Both of these subfields are encoded in the same manner, except that the last octet of the address field has the HDLC address extension bit set.

There is an octet at the end of each address subfield that contains the Secondary Station Identifier (SSID). The SSID subfield allows an Amateur Radio operator to have more than one packet-radio station operating under the same call sign. This is useful when an amateur wants to put up a repeater in addition to a regular station, for example. The C bits (see 2.4.1.2, below) and H bit (see 2.2.13.2, below) are also contained in this octet, along with two bits which are reserved for future use.

Fig. 3A shows a typical AX.25 frame in the nonrepeater mode of operation.

The frame shown is an I frame, not going through a level 2 repeater, from WB4JFI (SSID=0) to K8MMO (SSID=0), with no level 3 protocol. The P/F bit is set; the receive sequence number (N(R)) is 1; the send sequence number (N(S)) is 7.

2.2.13.1.1 Destination Subfield Encoding

Where:

1. The top octet (A1) is the first octet sent, with bit 0 of each octet being the first bit sent, and bit 7 being the last bit sent.
2. The first (low-order or bit 0) bit of each octet is the HDLC address extension bit, which is set to zero on all but the last octet in the address field, where it is set to one.
3. The bits marked "r" are reserved bits. They may be used in an agreed-upon manner in individual networks. When not implemented, they should be set to one.
4. The bit marked "C" is used as the command/response bit of an AX.25 frame, as outlined in 2.4.1.2 below.
5. The characters of the call sign should be standard seven-bit ASCII (upper case only) placed in the leftmost seven bits of the octet to make room for the address extension bit. If the call sign contains fewer than six characters, it should be padded with ASCII spaces between the last call sign character and the SSID octet.
6. The 0000 SSID is reserved for the first personal AX.25 station. This establishes one standard SSID for "normal" stations to use for the first station.

2.2.13.2 Level 2 Repeater-Address Encoding

In order to provide some method of indicating when a frame has been repeated, the H bit is set to zero on frames going to a repeater. The repeater will set the H bit to one when the frame is retransmitted. Stations should monitor the H bit, and discard any frames going to the repeater (uplink frames), while operating through a repeater. Fig. 4 shows how the repeater- address subfield is encoded. Fig. 4A is an example of a complete frame after being repeated.

1. The top octet is the first octet sent, with bit 0 being sent first and bit 7 sent last of each octet.
2. As with the source and destination address subfields discussed above, bit 0 of each octet is the HDLC address extension bit, which is set to zero on all but the last address octet, where it is set to one.
3. The "R" bits are reserved in the same manner as in the source and destination subfields.
4. The "H" bit is the has-been-repeated bit. It is set to zero whenever a frame has not been repeated, and set to one by the repeater when the frame has been repeated.

The above frame is the same as Fig. 3A, except for the addition of a repeater-address subfield (WB4JFI, SSID=1). The H bit is set, indicating this is from the output of the repeater.

2.2.13.3 Multiple Repeater Operation

The number of repeater addresses is variable. All but the last repeater address will have the address extension bits of all octets set to zero, as will all but the last octet (SSID octet) of the last repeater address. The last octet of the last repeater address will have the address extension bit set to one, indicating the end of the address field.

It should be noted that various timers (see 2.4.7, below) may have to be adjusted to accommodate the additional delays encountered when a frame must pass through a multiple-repeater chain, and the return acknowledgement must travel through the same path before reaching the source device.

It is anticipated that multiple-repeater operation is a temporary method of interconnecting stations over large distances until such time that a layer 3 protocol is in use. Once this layer 3 protocol becomes operational, repeater chaining should be phased out.

2.3 Elements of Procedure

2.3.1

2.3.2 Control-Field Formats and State Variables

2.3.2.1 Control-Field Formats

The control fields used in AX.25 use the CCITT X.25 control fields for balanced operation (LAPB), with an additional control field taken from ADCCP to allow connectionless and round- table operation.

There are three general types of AX.25 frames. They are the Information frame (I frame), the Supervisory frame (S frame), and the Unnumbered frame (U frame). Fig. 5 shows the basic format of the control field associated with these types of frames.

Where:

1. Bit 0 is the first bit sent and bit 7 is the last bit sent of the control field.
2. N(S) is the send sequence number (bit 1 is the LSB).
3. N(R) is the receive sequence number (bit 5 is the LSB).
4. The "S" bits are the supervisory function bits, and their encoding is discussed in 2.3.4.2.
5. The "M" bits are the unnumbered frame modifier bits and their encoding is discussed in 2.3.4.3.
6. The P/F bit is the Poll/Final bit. Its function is described in 2.3.3. The distinction between command and response, and therefore the distinction between P bit and F bit, is made by addressing rules discussed in 2.4.1.2.

2.3.2.1.1 Information-Transfer Format

2.3.2.1.2 Supervisory Format

2.3.2.1.3 Unnumbered Format

2.3.2.2 Control-Field Parameters

2.3.2.3 Sequence Numbers

2.3.2.4 Frame Variables and Sequence Numbers

2.3.2.4.1 Send State Variable V(S)

2.3.2.4.2 Send Sequence Number N(S)

2.3.2.4.3 Receive State Variable V(R)

2.3.2.4.4 Received Sequence Number N(R)

2.3.3 Functions of Poll/Final (P/F) Bit

2.3.4 Control Field Coding for Commands and Responses

2.3.4.1 Information Command Frame Control Field

The information-frame control field is encoded as shown in Fig. 6. These frames are sequentially numbered by the N(S) subfield to maintain control of their passage over the link-layer connection.

2.3.4.2 Supervisory Frame Control Field

2.3.4.2.1 Receive Ready (RR) Command and Response

1. to indicate that the sender of the RR is now able to receive more I frames.
2. to acknowledge properly received I frames up to, and including N(R)-1, and
3. to clear a previously set busy condition created by an RNR command having been sent.

The status of the DXE at the other end of the link can be requested by sending a RR command frame with the P-bit set to one.

2.3.4.2.2 Receive Not Ready (RNR) Command and Response

The RNR condition can be cleared by the sending of a UA, RR, REJ, or SABM frame.

The status of the DXE at the other end of the link can be requested by sending a RNR command frame with the P bit set to one.

2.3.4.2.3 Reject (REJ) Command and Response

Only one reject frame condition is allowed in each direction at a time. The reject condition is cleared by the proper reception of I frames up to the I frame that caused the reject condition to be initiated.

The status of the DXE at the other end of the link can be requested by sending a REJ command frame with the P bit set to one.

2.3.4.3 Unnumbered Frame Control Fields

Fig. 8 shows the layout of U frames implemented within this protocol.

2.3.4.3.1 Set Asynchronous Balanced Mode (SABM) Command

Information fields are not allowed in SABM commands. Any outstanding I frames left when the SABM command is issued will remain unacknowledged.

The DXE confirms reception and acceptance of a SABM command by sending a UA response frame at the earliest opportunity. If the DXE is not capable of accepting a SABM command, it should respond with a DM frame if possible.

2.3.4.3.2 Disconnect (DISC) Command

Prior to acting on the DISC frame, the receiving DXE confirms acceptance of the DISC by issuing a UA response frame at its earliest opportunity. The DXE sending the DISC enters the disconnected state when it receives the UA response.

Any unacknowledged I frames left when this command is acted upon will remain unacknowledged.

2.3.4.3.3 Frame Reject (FRMR) Response

2.3.4.3.3.1

The FRMR response frame is sent to report that the receiver of a frame cannot successfully process that frame and that the error condition is not correctable by sending the offending frame again. Typically this condition will appear when a frame without an FCS error has been received with one of the following conditions:

1. The reception of an invalid or not implemented command or response frame.
2. The reception of an I frame whose information field exceeds the agreed-upon length. (See 2.4.7.3, below.)
3. The reception of an improper N(R). This usually happens when the N(R) frame has already been sent and acknowledged, or when N(R) is out of sequence with what was expected.
4. The reception of a frame with an information field where one is not allowed, or the reception of a U or S frame whose length is incorrect. Bits W and Y described in 2.3.4.3.3.2 should both be set to one to indicate this condition.
5. The reception of a supervisory frame with the F bit set to one, except during a timer recovery condition (see 2.4.4.9), or except as a reply to a command frame sent with the P bit set to one. Bit W (described in 2.3.4.3.3.2) should be set to one.
6. The reception of an unexpected UA or DM response frame. Bit W should be set to one.
7. The reception of a frame with an invalid N(S). Bit W should be set to one.

An invalid N(R) is defined as one which points to an I frame that previously has been transmitted and acknowledged, or an I frame which has not been transmitted and is not the next sequential I frame pending transmission.

An invalid N(S) is defined as an N(S) that is equal to the last transmitted N(R)+k and is equal to the received state variable V(R), where k is the maximum number of outstanding information frames as defined in 2.4.7.4 below.

An invalid or not implemented command or response is defined as a frame with a control field that is unknown to the receiver of this frame.

2.3.4.3.3.2

Where:

1. The rejected frame control field carries the control field of the frame that caused the reject condition. It is in bits 0-7 of the information field.
2. V(S) is the current send state variable of the device reporting the rejection (bit 9 is the low bit).
3. The CR bit is set to zero to indicate the rejected frame was a command, or one if it was a response.
4. V(R) is the current receive state variable of the device reporting rejection (bit 13 is the low bit).
5. If W is set to 1, the control field received was invalid or not implemented.
6. If X is set to 1, the frame that caused the reject condition was considered invalid because it was a U or S frame that had an information field that is not allowed. Bit W must be set to 1 in addition to the X bit.
7. If Y is set to 1, the control field received and returned in bits exceeded the maximum allowed under this recommendation in 2.4.7.3, below.
8. If A is set to 1, the control field received and returned in bits 1 to 8 contained an invalid N(R).
9. Bits 8, and 20 to 23 are set to 0.

2.3.4.3.4 Unnumbered Acknowledge (UA) Response

2.3.4.3.5 Disconnected Mode (DM) Response

Whenever a SABM frame is a received, and it is determined that a connection is not possible, a DM frame shall be sent. This will indicate that the called station cannot accept a connection at that time.

While a DXE is in the disconnected mode, it will respond to any command other than a SABM or UI frame with a DM response with the P/F bit set to 1.

2.3.4.3.6 Unnumbered Information (UI) Frame

The Unnumbered Information frame contains PID and information fields and is used to pass information along the link outside the normal information controls. This allows information fields to go back and forth on the link bypassing flow control. Since these frames are not acknowledgeable, if one gets obliterated, there is no way to recover it. A received UI frame with the P bit set shall cause a response to be transmitted. This response shall be a DM frame when in the disconnected state or a RR (or RNR, if appropriate) frame in the information transfer state.

2.3.5 Link Error Reporting and Recovery

2.3.5.1 DXE Busy Condition

2.3.5.2 Send Sequence Number Error

The control field of the erroneous I frame(s) will be accepted so that link supervisory functions such as checking the P/F bit can still be performed. Because of this updating, the retransmitted I frame may have an updated P bit and N(R).

2.3.5.3 Reject (REJ) Recovery

A DXE receiving the REJ command will clear the condition by resending all outstanding I frames (up to the window size), starting with the one indicated in N(R) of the REJ frame.

2.3.5.4 Time-out Error Recovery

2.3.5.4.1 T1 Timer Recovery

2.3.5.4.2 Timer T3 Recovery

2.3.5.5 Invalid Frame or FCS Error

2.3.5.6 Frame Rejection Condition

Once a rejection error occurs, no more I frames are accepted (except for the examination of the P/F bit) until the error is resolved. The error condition is reported to the other DXE by sending a FRMR response frame. See 2.4.5.

2.4 Description of AX.25 Procedures

2.4.1 Address Field Operation

2.4.1.1 Address Information

The destination address can be a group name or club call sign if the point-to-multipoint operation is allowed. Operation with destination addresses other than actual amateur call signs is a subject for further study.

2.4.1.2 Command/Response Procedure

An upward-compatible AX.25 DXE can determine whether it is communicating with a DXE using an older version of this protocol by testing the command/response bit information located in bit 7 of the SSID octets of both the destination and source address subfields. If both C bits are set to zero, the device is using the older protocol. The newer version of the protocol always has one of these two bits set to one and the other set to zero, depending on whether the frame is a command or a response.

The command/response information is encoded into the address field as shown in Fig. 10.

Since all frames are considered either commands or responses, a device shall always have one of the bits set to one, and the other bit set to zero.

The use of the command/response information in AX.25 allows S frames to be either commands or responses. This aids maintenance of proper control over the link during the information transfer state.

2.4.2 P/F Bit Procedures

The next response frame returned to an I frame with the P bit set to 1, received during the information transfer state, will be a RR, RNR, or REJ response with the F bit set to 1.

The next response frame returned to a supervisory command frame with the P bit set to 1, received during the information transfer state, will be a RR, RNR, or REJ response frame with the F bit set to 1.

The next response frame returned to a S or I command frame with the P bit set to 1, received in the disconnected state, will be a DM response frame with the F bit set to 1.

The P bit is used in conjunction with the time-out recovery condition discussed in 2.3.5.4, above.

When not used, the P/F bit is set to zero.

2.4.3 Procedures For Link Set-Up and Disconnection

2.4.3.1 LAPB Link Connection Establishment

If the other DXE doesn't respond before T1 times out, the device requesting the connection will re-send the SABM frame, and start T1 running again. The DXE should continue to try to establish a connection until it has tried unsuccessfully N2 times. N2 is defined in 2.4.7.2, below.

If, upon reception of a SABM command, the DXE decides that it cannot enter the indicated state, it should send a DM frame.

When receiving a DM response, the DXE sending the SABM should cancel its T1 timer, and not enter the information- transfer state.

The DXE sending a SABM command will ignore and discard any frames except SABM, DISC, UA, and DM frames from the other DXE.

Frames other than UA and DM in response to a received SABM will be sent only after the link is set up and if no outstanding SABM exists.

2.4.3.2 Information-Transfer Phase

When receiving a SABM command while in the information- transfer state, the DXE will follow the resetting procedure outlined in 2.4.6 below.

2.4.3.3 Link Disconnection

2.4.3.3.1

2.4.3.3.2

2.4.3.3.3

2.4.3.4 Disconnected State

2.4.3.4.1

2.4.3.4.2

2.4.3.4.3

2.4.3.4.4

2.4.3.5 Collision Recovery

2.4.3.5.1 Collisions in a Half-Duplex Environment

2.4.3.5.2 Collisions of Unnumbered Commands

If sent and received SABM or DISC commands are different, both DXEs should enter the disconnected state and transmit a DM frame at the earliest opportunity.

2.4.3.5.3 Collision of a DM with a SABM or DISC

2.4.3.6 Connectionless Operation

The way round-table activity is implemented is technically outside the AX.25 connection, but still using the AX.25 frame structure.

AX.25 uses a special frame for this operation, called the Unnumbered Information (UI) frame. When this type of operation is used, the destination address should have a code word installed in it to prevent the users of that particular round table from seeing all frames going through the shared RF medium. An example of this is if a group of amateurs are in a round-table discussion about packet radio, they could put PACKET in the destination address, so they would receive frames only from others in the same discussion. An added advantage of the use of AX.25 in this manner is that the source of each frame is in the source address subfield, so software could be written to automatically display who is making what comments.

Since this mode is connectionless, there will be no requests for retransmissions of bad frames. Collisions will also occur, with the potential of losing the frames that collided.

2.4.4 Procedures for Information Transfer

2.4.4.1 Sending I Frames

The DXE should not transmit any more I frames if its send state variable equals the last received N(R) from the other side of the link plus seven. If it were to send more I frames, the flow control window would be exceed, and errors could result.

If a DXE is in a busy condition, it may still send I frames as long as the other device is not also busy.

If a DXE is in the frame-rejection mode, it will stop sending I frames.

2.4.4.2 Receiving I Frames

2.4.4.2.1

1. If it has an I frame to send, that I frame may be sent with the transmitted N(R) equal to its receive state variable V(R) (thus acknowledging the received frame). Alternately, the device may send a RR frame with N(R) equal to V(R), and then send the I frame.
2. If there are no outstanding I frames, the receiving device will send a RR frame with N(R) equal to V(R). The receiving DXE may wait a small period of time before sending the RR frame to be sure additional I frames are not being transmitted.

2.4.4.2.2

If a busy condition exists, the DXE receiving the busy condition indication should poll the sender of the busy indication periodically until the busy condition disappears.

A DXE may poll the busy DXE periodically with RR or RNR frames with the P bit set to one.

The reception of I frames that contain zero-length information fields shall be reported to the next level but no information field will be transferred.

2.4.4.3 Reception of Out of Sequence Frames

2.4.4.4 Reception of Incorrect Frames

2.4.4.5 Receiving Acknowledgement

2.4.4.6 Receiving Reject

1. If the DXE is not transmitting at the time, and the channel is open, the device may commence to retransmit the I frame(s) immediately.
2. If the DXE is operating on a full-duplex channel transmitting a UI or S frame when it receives a REJ frame, it may finish sending the UI or S frame and then retransmit the I frame(s).
3. If the DXE is operating in a full-duplex channel transmitting another I frame when it receives a REJ frame, it may abort the I frame it was sending and start retransmission of the requested I frames immediately.
4. The DXE may send just the one I frame outstanding, or it may send more than the one indicated if more I frames followed the first one not acknowledged, provided the total to be sent does not exceed the flow-control window (7 frames).

If the DXE receives a REJ frame with the poll bit set, it should respond with either a RR or RNR frame with the final bit set before retransmitting the outstanding I frame(s).

2.4.4.7 Receiving a RNR Frame

2.4.4.8 Sending a Busy Indication

2.4.4.9 Waiting Acknowledgement

If timer T1 runs out before a supervisory response frame with the F bit set is received, the DXE will retransmit an appropriate supervisory command frame (RR or RNR) with the P bit set. After N2 attempts to get a supervisory response frame with the F bit set from the other DXE, the DXE will initiate a link resetting procedure as described in 2.4.6, below.

2.4.5 Frame Rejection Conditions

Under these conditions, the DXE will ask the other DXE to reset the link by transmitting a FRMR response as outlined in 2.4.6.3, below.

After sending the FRMR frame, the sending DXE will enter the frame reject condition. This condition is cleared when the DXE that sent the FRMR frame receives a SABM or DISC command, or a DM response frame. Any other command received while the DXE is in the frame reject state will cause another FRMR to be sent out with the same information field as originally sent.

In the frame rejection condition, additional I frames will not be transmitted, and received I frames and S frames will be discarded by the DXE.

The DXE that sent the FRMR frame shall start the T1 timer when the FRMR is sent. If no SABM or DISC frame is received before the timer runs out, the FRMR frame shall be retransmitted, and the T1 timer restarted as described in the waiting acknowledgement section (2.4.4.9) above. If the FRMR is sent N2 times without success, the link shall be reset.

2.4.6 Resetting Procedure

2.4.6.1

2.4.6.2

2.4.6.3

If a DM response is received, the DXE will enter the disconnected state and stop timer T1. If timer T1 runs out before a UA or DM response frame is received, the SABM will be retransmitted and timer T1 restarted. If timer T1 runs out N2 times, the DXE will enter the disconnected state, and any previously existing link conditions will be cleared.

Other commands or responses received by the DXE before completion of the reset procedure will be discarded.

2.4.6.4

2.4.7 List of System Defined Parameters

2.4.7.1 Timers

2.4.7.1.1 Acknowledgement Timer T1

If level 2 repeaters are to be used, the value of T1 should be adjusted according to the number of repeaters the frame is being transferred through.

2.4.7.1.2 Response Delay Timer T2

2.4.7.1.3 Inactive Link Timer T3

2.4.7.2 Maximum Number of Retries (N2)

2.4.7.3 Maximum Number of Octets in an I Field (N1)

2.4.7.4 Maximum Number of I Frames Outstanding (k)