5
$\begingroup$

I've recently been reading a number of articles on the current aviation systems. Many of them say that the current aviation infrastructure is 'inefficient' since it is based on OSI model and, so, Eurocontrol/FAA are trying to shift to a TCP/IP model (without further elaboration as to why it is inefficient.)

As far as I know OSI and TCP/IP are both protocol stacks and pretty much follow the same ideas.

I'm not quite sure why they say that the current OSI model is inefficient? In what sense is it inefficient?

Another point, in the articles I read, which I thought was a bit vague, was that it said that the TCP/IP model is more 'modular' and can improve 'flight delays'.

Again, weren't both OSI and TCP/IP models created to make systems more modular? How can shifting to TCP/IP make current aviation systems more delay-tolerant given that the main cause of delay is air traffic and weather?

Articles I've read.

  1. This one is the main article, published by Eurocontrol/FAA, that I read. It suggests a transition from a voice-based to data-based services network.

  2. This one is a PhD thesis. See Section 3.1 for the main references. They refer to the TCP/IP ATN as ATN/IPS. Search for the word 'ATN/IPS'.

  3. This one gives a detailed description of ATN/IPS.

  4. The following claims that a TCP/IP model is more modular than the current OSI based model.

$\endgroup$
7
  • 4
    $\begingroup$ Could you post a link to this article? $\endgroup$
    – GdD
    Commented Oct 17, 2017 at 14:08
  • 5
    $\begingroup$ On a side note, TCP/IP is one of the layers (4) in the OSI model, so I'm not quite sure what their analogy is trying to say without reading the article itself. Unless they are referring to this (DARPA model), which is essentially the same as the OSI model - technet.microsoft.com/en-us/library/cc958821.aspx $\endgroup$ Commented Oct 17, 2017 at 14:12
  • 8
    $\begingroup$ The OSI model is a generic conceptual model for understanding data movement in telecommunications systems. It's an educational tool, not something you could implement. This question doesn't make sense I'm afraid. $\endgroup$
    – GdD
    Commented Oct 17, 2017 at 14:16
  • 1
    $\begingroup$ eurocontrol.int/sites/default/files/field_tabs/content/… $\endgroup$
    – Teodorism
    Commented Oct 17, 2017 at 14:16
  • 1
    $\begingroup$ Neither OSI nor TCP/IP are mentioned in the report you linked to in your comment, at least they aren't in the abbreviations list and the Find function doesn't find them. Can you edit your question to add a link to the report and some actual quotes from it to help explain what you're reading? $\endgroup$
    – Pondlife
    Commented Oct 17, 2017 at 14:23

1 Answer 1

11
$\begingroup$

The current air/ground systems (based on OSI) are aviation unique. And they are limited by design to "messages no larger than 3.5 kilobytes". This limits how much information can be sent to/from an airplane, and is a major roadblock towards time based navigation and the broader data-hungry Global Air Navigation Plan (GANP), which includes SWIM for example.

SWIM: an advanced technology program designed to facilitate greater sharing of Air Traffic Management (ATM) system information, such as airport operational status, weather information, flight data, status of special use airspace, and National Airspace System (NAS) restrictions.

Option one, develop better aviation-unique systems (OSI) that can handle more data. Very expensive to certify, very hard to get everyone to agree on the system. Option two, use the IPS (Internet Protocol Suite) that has proven itself. And thus commercial off-the-shelf equipment can be used.

The reasoning at the time was that the OSI protocols were more formally specified and included such things as detailed protocol implementation conformance matrices and combinations of functions at each protocol level called profiles, and thus were more suitable for avionics certification.

(...)

However, as is common knowledge, the TCP/IP protocols have long since been proven in the World Wide Web and therefore are able to offer potential economic benefit and provide for the rapid introduction of new services in the global aviation environment.

  • OSI means OSI-based (such as ARINC 429), and is unique to aviation. Example: A router on a Boeing 777 is not sold on the non-aviation market. What they want to do is use off-the-shelf solutions to be used on airplanes. Using ready-made solutions is more economical than creating solutions for a specific industry. The current solutions can't handle the data requirements of the future.

IPS will also allow easier integration with the already IPS based ground/ground communication between airlines and air traffic management. The current plan is to have a seamless transition from OSI to OSI+IPS to IPS in the next 15 years or so.

Why is IPS more delay-tolerant?

Example: When every single user (pilot, airline, ATC, airport, etc.) has access to the same real-time weather data (e.g. via SWIM), decisions can be made more correctly, in a timely manner, and consistently across the users. Just like what FANS-1 did for the Pacific operations in the 90's, but way bigger (history below).

This is the gist of it and explains why the shift. For technical standards there are plenty of ICAO documents on the web: ICAO Doc 9705 (OSI) and ICAO Doc 9896 (IPS).

Also related: Why Boeing and Airbus are Pushing for IPS (aviationtoday.com)


A brief history of aviation data comms:

  • ACARS started as a punch clock for the flight crews by auto reporting the different flight phases (aka OOOI)
  • Then Europe expanded on what the US has done with it and used it for operational control, e.g., by keeping track of where the fleets are, etc. (ops driven)
  • Then British Airways took it a step further for airplane health monitoring (engineering driven)
  • Then the Pacific region wanted to benefit from it, and so came Boeing's FANS-1, enabling flights that took off with +11 hours old weather forecasts to readjust their routes mid-flight. The flights can now receive from the airlines the weather updates, new flex-routes, etc., and communicate it to the ATC. Shaving off 15+ minutes on long transpacific flights, and enabling more payload (2+ tonnes per flight)
  • Now FANS serves ATC (CPDLC and ADS-C) in oceanic airspace, and operators (AOC)
  • Airbus also made its own FANS-A, now they are known as FANS-1/A
  • Then they became FANS-1/A+ by adding a latency timer for the ATC messages (CPDLC), so a plane would not receive a delayed message that was meant for a previous flight this plane flew
  • Then PM-CPDLC (aka LINK2000+ and ATN-B1) came along in Europe, which requires yet again different equipment.

For ATC, FANS-1/A+ is mainly used in oceanic airspace and in delivery clearances in the US, and CPDLC (ATN-B1) in parts of the European airspace, and the hardware for both is not compatible (if used together) on older planes.

FANS-1/A+ combined with ATN-B1 creates FANS-2/B as one package (like on the Boeing 787). The plan is to make all systems use a standard network with higher bandwidth, which is the current goal of FANS-3/C (FANS-1/A+ and ATN-B2) using an IPS infrastructure while supporting the legacy planes. Check the progression of the 3 stars I added below.

enter image description here
(ICAO)

  • Note: the acronyms above may seem to overlap, e.g., for ACARS and CPDLC, but as the uses expanded and new systems were added, the meaning now depends on the context. For example the early FANS-1 CPDLC is still ACARS at its core, and also CPDLC can refer to ATN-B1 in Europe, or DCL-CPDLC (DCL = Delivery Clearance) in the US for example. It's an alphabet soup.

Further reading:

$\endgroup$
2
  • $\begingroup$ I'm curious to know, are there potential (control and operational) services that ATN/IPS can offer? or is ATN/IPS intended for cost efficiency and latency reasons only. $\endgroup$
    – Teodorism
    Commented Oct 19, 2017 at 16:21
  • $\begingroup$ Look up the Global Air Navigation Plan (GANP) I mentioned and presented a part of it: SWIM. $\endgroup$
    – user14897
    Commented Oct 19, 2017 at 18:33

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .