[ADS-L](https://skyblip.eu/ads-l.md)

# Our open questions on ADS-L Issue 2.

These are the questions skyBlip puts to EASA on ADS-L 4 SRD-860 Issue 2, where the text leaves a reading open. Each theme quotes the clause, records what our firmware does today and says whether the question has been sent and answered. Until EASA answers, Go flies the reading described here.

skyBlip is open-source electronic conspicuity equipment for general aviation. skyBlip Go transmits and receives ADS-L. skyBlip Bravo receives ADS-B on 1090 MHz and transmits nothing. Paired with a Go, on the ground, Bravo can name the transponder it takes for the aircraft’s own.

## The ICAO address

### Which address should Go send, and which settings must it refuse?

Clause **ADS-L.4.SRD860.F.2.3, “ICAO Address”** , in [ADS-L 4 SRD-860, ED Decision 2022/024/R, Issue 2](https://www.easa.europa.eu/en/document-library/agency-decisions/ed-decision-2022024r), says:

> For this setting, the ICAO address shall be configurable and entered by the user of the ADS-L system, thereby overriding the address given by the manufacturer. This is the preferred mode for aircraft that have an ICAO address (Mode-S transponder code). This setting shall be revertible by user configuration. For aircraft that concurrently transmit using other e-conspicuity systems such as ADS-B, or aircraft that use a Mode-S transponder, it shall use this type of address to allow for disambiguation and attribution of the different radio transmissions on the receiver side. The ADS-L system shall enforce this by disallowing inconsistent user configurations.

Today Go uses the pilot’s configured ICAO address on every payload and refuses an address no aircraft holds (`FFFFFF`) or an entered address that conflicts with the transponder Bravo identifies on the ground.

1. **Should every payload use the ICAO address?** When the pilot enters the aircraft’s ICAO address, Go sends every payload under Address Mapping Table (AMT) entry 5, ICAO, with that address. This includes Traffic, Status and payload type 66, which carries the aircraft’s registration.

   Go no longer transmits under skyBlip’s manufacturer address: AMT entry 58, manufacturer page 49, prefix 0, with the unit’s serial in the remaining 16 bits. The ADS-L registry assigned that range to skyBlip. Clearing the ICAO setting restores AMT entry 58 and the unit’s address from the next burst. Is this the intent of F.2.3, or should some payloads stay under the manufacturer’s address?

2. **What must “disallowing inconsistent user configurations” cover?** Go refuses `FFFFFF`, an address no aircraft holds. On the ground, while a paired Bravo names the transponder it takes for the aircraft’s own, Go also refuses an entered address that differs from it.

   Go does not refuse an unset ICAO address when Bravo identifies a Mode-S transponder on the aircraft. It proposes the address Bravo hears and lets the pilot accept it. Does F.2.3 require Go to refuse that configuration, or to enforce the address itself? Are there other configurations EASA intends the device to disallow?

Asked: [asked on date]. No answer from EASA yet.

## The flight state

### Should a craft that can hover announce Undefined rather than On ground?

Clause **ADS-L.4.SRD860.G.1.2, “Flight State”** , says:

> The “Undefined” value shall be used if no reasonable decision is possible by the ADS-L device.

Go decides the flight state from ground speed. A paraglider soaring a ridge into wind, a balloon or a hovering helicopter can fly with no ground speed, so for those aircraft categories Go cannot tell a landing from flight. For categories 3, 5, 7 to 13 and 15 to 17 it sends Undefined instead of On ground. It therefore keeps the 1 Hz rate of G.1.16 instead of dropping to the ground rate of 0.1 Hz.

1. **Is this the use of Undefined the clause intends?** Or should these categories announce On ground whenever their ground speed is low, and transmit at the ground rate?

Not sent yet.

## The callsign

### May payload type 66 go out after the direct slot?

Clause **ADS-L.4.SRD860.C.5** gives aircraft the direct slot, from 450 to 1000 ms of each UTC second, and says of the rest of the second:

> Reserved: The unused time is reserved for future extensions and shall not be used for any ADS-L transmissions.

Clause **ADS-L.4.SRD860.F.2.1, “Payload Type Identifier”** , lists payload type 66 as “OGN Diagnostics” and says its “Payload definition is outside the scope of this document.”

Go sends its callsign in OGN’s payload type 66 registration record, on the ADS-L frame, between 0 and 200 ms of the second after its slot. The [callsign page](https://skyblip.eu/ads-l/callsign.md) explains that record.

1. **Is a payload the specification does not define an ADS-L transmission in the sense of C.5?** We read it as not being one. Or does payload type 66 belong in the direct slot?

Not sent yet. The callsign lives outside Go’s ADS-L stack, so a different answer changes the callsign code and not the ADS-L version.

## The age of the position

### Must every position be a solution measured on a quarter second and sent within 500 ms?

Clause **ADS-L.4.SRD860.G.1.1, “Timestamp”** , says:

> The Timestamp field indicates the number of quarter seconds since the full hour, modulo 60. […] The timestamp refers to the time of the navigation solution (PVT fix) obtained by the GNSS receiver (e.g. on the full second) and included in the Traffic payload. It shall be accurate to ± 10 ms.

Clause **ADS-L.4.SRD860.G.1.16, “Transmit Rate and Timing”** , says:

> A message shall be replaced with a new message if the total delay of 500 ms from obtaining (measuring) the PVT fix is exceeded, e.g. due to the frequency being occupied, to prevent old data being sent. No Traffic message shall be sent containing navigation data older than 500 ms.

Read together with the direct slot of C.5, from 450 to 1000 ms, every position must be a solution the receiver measured within 10 ms of a quarter second, and it must leave within 500 ms of that measurement. Go does not move a position forward to a quarter second, because the result would no longer be a solution the receiver obtained.

This ties each transmission to the receiver’s solution schedule, which we have not met in other systems. A receiver solving once a second, on the full second, serves only bursts between 450 and 500 ms: 50 ms of the 550 ms slot. At 5 solutions a second, only the full second falls on a quarter. Only 4 solutions a second, or a multiple, puts every solution on a quarter. Go’s receiver, a Quectel L76K, documents 1, 2 and 5 solutions a second, and on the bench it ignored a request for 4. Go therefore runs it at 2, with solutions on .000 and .500 of each second.

At 2 solutions a second, part of the slot has no usable solution. Go sends on 868.2 MHz between 450 and 810 ms and on 868.4 MHz between 830 and 990 ms ([timing page](https://skyblip.eu/timing.md)). The .000 solution serves bursts up to 490 ms, keeping 10 ms for the burst’s air time. The .500 solution reaches Go 40 to 160 ms after it was measured. Between 490 ms and its arrival, no solution meets both clauses. That gap falls inside the 868.2 MHz window. In 26 minutes on the bench, before the change below, 21 of the 81 positions due on 868.2 MHz were not sent.

Go now holds a position that falls in the gap until the .500 solution arrives. If the position’s instant has passed by then, Go draws a new instant at random over the rest of the window, from its address and the second, so units do not all transmit on the solution’s arrival. A position is lost only if the .500 solution arrives after 810 ms, the window’s last instant. This is not ideal. On 868.2 MHz the bursts no longer spread evenly: about a quarter move from 490–590 ms into 590–810 ms, which then carries about 40 % more bursts, and the chance of two units colliding rises from about 2.8 % to about 3.9 % per neighbour. Every Go has the same receiver and the same gap, so the uneven load adds up across a fleet. It also depends on the receiver’s delay, which Go measures but does not control.

1. **Do G.1.1 and G.1.16 intend to bind each position to a solution measured on a quarter second?** Or may a device carry its latest solution forward to the quarter the Timestamp names, with the 500 ms counted from that measurement?

2. **Could a later issue let the Timestamp name the solution’s own time more finely than a quarter second?** Receivers solving 1, 5 or 10 times a second could then meet both clauses, and devices could keep an even spread over the slot.

Not sent yet.

## Channel alternation

### Does alternation by the schedule meet the rule after a burst that did not go out?

Clause **ADS-L.4.SRD860.C.2.5, “Frequency Diversity”** , says:

> For the “Traffic” and “Status” payloads (see Subpart G), the sender shall alternate between these frequencies for every transmission to facilitate reception of these payloads for restricted, single-channel receivers.

Go picks the channel from the parity of the UTC second: a position on 868.2 MHz in an even second and on 868.4 MHz in an odd one, and the Status the other way round. When a burst does not go out, because the channel stayed busy, the next one keeps its own second’s channel, so two positions in a row can leave on the same frequency. A takeoff or a landing changes the rate, and in about one case in two it puts two positions in a row on one channel, once. A single-channel receiver still hears Go every other second, which is the purpose the clause gives.

1. **Does alternation by the schedule meet C.2.5?** Or must each transmission take the channel opposite to the last one that actually went out?

Not sent yet.

## Velocity accuracy

### May a device with no measured velocity bound send it as unknown?

Clause **ADS-L.4.SRD860.G.1.17, “Quality Indicators”** , says:

> Unlike the ADS-B certification approach (CS-STAN Configuration 3 in particular), uncertified ADS-L devices shall however not report all quality values as 0. Instead, they shall report their positioning performance dynamically as defined in the previous sections.

Go’s receiver reports its horizontal and vertical dilution of precision, but no velocity error. Go’s horizontal and vertical accuracy change with each fix, while its velocity accuracy is sent as 0, “Unknown”, which the table of G.1.16, “Velocity Accuracy”, allows.

1. **Does that meet G.1.17?** Not every quality value is 0, but the velocity accuracy always is.

Not sent yet.

## References

- **EASA** · [ADS-L 4 SRD-860](https://www.easa.europa.eu/en/document-library/agency-decisions/ed-decision-2022024r), ED Decision 2022/024/R, Issue 2. C.2.5, C.5, F.2.1, F.2.3, G.1.1, G.1.2, G.1.16 and G.1.17, quoted above.
- **skyBlip firmware** · [`core/settings/README.md`](https://github.com/fcatuhe/skyblip/blob/main/firmware/core/settings/README.md). The ICAO address section records the setting, the configurations Go refuses and the rules considered but not imposed.
- **skyBlip firmware** · [`docs/ADSL_EASA_QUESTIONS.md`](https://github.com/fcatuhe/skyblip/blob/main/docs/ADSL_EASA_QUESTIONS.md). The same questions, with the firmware files behind each reading.

[ADS-L NMEA proposal for Bluetooth displays](https://skyblip.eu/ads-l/nmea.md)
