Provider Abort Action Plan

From WikiLink
Jump to: navigation, search

Provider Aborts (PAs) are a major concern. They occur when there is a sustained loss of end-to-end connectivity. The problems causing the loss of connectivity may arise anywhere within CPDLC; the ATN and VDL Mode 2. There is no single cause for all PAs. Provider Aborts have been occuring for several years and over that time a number of problems have been identified and fixed. A major cause of PAs is a phenomenon known as 'VDR Deafness' where the aircraft continues to downlink messages normally but fails to recognise uplinks. This problem occurs with Rockwell Collins radios and a fix is now available. However 'VDR Deafness' is only one cause of PAs (albeit a major one) and the rate of PAs even when the VDR deafness update is fully deployed will still remain much higher than is desirable.

Starting with an ad-hoc meeting in May 2013 a PA action plan was developed and a number of investigations performed. The list below represents the plan and its status around the end of 2014. Since that time the SJU established the ELSA project to investigate the causes of the current PA problems and the ELSA project now provides the main thrust of the investigations, although the CRO also continues to perform some investigations.

CRO led investigations

  • Individual PAs for Aircraft with VDR Deafness update installed. The CRO have analysed in detail the causes of a large sample (69) of the PAs suffered by a specific set of aircraft that have update to address VDR deafness. This analysis has identified a number of problems principally with the communications network which are currently under further investigation. The most significant issue uncovered (a factor in approx. 50% of these PAs) was the failure of the uplink i.e. the maximum number of N2 retries being exceeded and further work investigating the causes of this is planned (see below).
  • Long Delay Investigations. To investigate transactions suffering from significant network delays (but that don’t result in a PA). The CRO have also analysed a sample (32) of transactions for which the technical round trip delay exceeds 40 seconds for a set of aircraft with the VDR deafness update installed. This analysis has not highlighted any new issues that were not raised by analysing the PAs suffered by the same set of aircraft.
  • N2 retries investigations. One of the factors behind approximately 50% of the PAs analysed from aircraft with the VDR deafness update and also a major source of the delays seen from the analysis of messages with long delay is the fact that the uplink fails at the AVLC layer i.e. the frame is transmitted N2 times without success . In order to investigate what might be causing these N2 retries the following initial steps have been taken:
  1. Based on VGS logs for 24 hours for a single VGS showing all traffic establish the distribution of retries for frames carrying ATN traffic and then check whether the number of N2 failures seen is simply representing the tail of a statistical distribution i.e. whether there is a specific trigger that is causing these uplink problems or whether it is a more general problem with the uplink performance. This work has been completed. It showed that there are more N2 events than would be expected which implies it is not simply an affect of congestion on the channel.
  2. Compare VGS logs with MOON logs to be sure that the frames are really being transmitted and it’s not a bug in the VGS. This work has been completed and showed no significant difference between the VGS logs and the MOON logs.
  3. Investigate the effect of AVLC frame length. Obviously the longer a frame is the more likely it is to get corrupted during transmission. Several days worth of VGS logs have been analysed and there is no evidence of an association between the length of the frame and the probability of an N2 event. Analysis of this data also showed no evidence that particular avionics are causing the N2 events, nor that signal strength plays a role.
  • Aircraft Cosite Interference. To study the impact of co-site interference between the airlines OPC voice channels and the datalink channels and whether this contributes to PAs it is proposed to identify some airlines which have OPC channels closest to the CSC and which experience PAs (ideally high rate and ideally close frequencies) and gather data about when the OPC channel is being used (not the content), on what channel and for what duration for an agreed set of aircraft over an agreed period. Then use the data collected to analyse any relationship between OPC use and PAs. This work has not started and is likely to be overtaken by the SJU study that is expected to start in January 2015.
  • FINNAIR High PA Rate versus DLH. It has been observed over the past years that FINNAIR suffer a much higher PA rate than similar DLH aircraft (same aircraft type, same avionics). The reasons for this are unknown; one possibility is simply that the FINNAIR aircraft suffer more from VDR deafness, so to determine whether this is true, the CRO plan to monitor the PA performance of some FINNAIR aircraft and some DLH aircraft fitted with the VDR deafness update and compare the PA rates of each set. This comparison was last attempted in February 2014 but at that time there was insufficient data available from aircraft fitted with the VDR deafness update to draw any conclusions. The plan is to repeat this analysis once more data is available.
  • Geographic Distribution Investigations. To investigate any geographic factors that may affect the rate at which aircraft suffer PAs. The CRO has created a series of heat maps (using google earth) that show in which areas a higher than average PA rate occurs. A separate map has been created for SITA airlines, ARINC airlines and one for aircraft that have implemented the fix for VDR deafness. These maps have been made available to ARINC and SITA for their consideration and analysis and are available to the broader datalink community on the DLISG One sky team.

Stakeholder led investigations

  • Honeywell Investigations. Honeywell are investigating the problems found with the EPIC/Embraer combination. They have gathered data from an instrumented aircraft flying in March 2014 and are currently analysing the data. They are also analysing data gathered from a set of KLM flights in the summer 2014.
  • Airbus In Service Evaluations. Airbus have conducted in service evaluations of the ATSU CSB 8.2 with favourable results.