The E-TDMA concept - Towards a new VDL strategy: Some key issues

1
Towards a new VDL strategy Some key issues
possible way forward
2

• Why a new VDL strategy ?
• no requirement for integrated voice-data
• the main datalink activity today is AOC
• the development of ATSC will be slow
• performance requirements are still unclear
• efficiency requires a data integration strategy
• the technology-driven approach is flawed
• Neither VDL Mode 3 nor VDL Mode 4 are adequate.
• A more flexible second generation system is
  needed
• to provide an efficient general purpose datalink

3

• The E-TDMA design approach
• Identify requirements and constraints for
  designing a general purpose VDL TDMA
• derive a stable list of design drivers
• Propose generic solutions that can be tuned to a
  variety of operational conditions and
  quantitative QoS requirements
• Discuss them with CAAs and manufacturers

4
Key issues for a general purpose VDL TDMA (1)
provide a sustainable migration path (take into
account the initially limited number of available
channels) (2) limit the cost of aircraft
equipment (stick to no more than 3 multi-purpose
VDR, one for the voicelink, one for the
datalink, and one as backup for either voicelink
or datalink, and avoiding an awkward
multiplication of datalink emittors) (3) support
end-to-end safety certification (offer a
deterministic behaviour and traceable QoS
specifications, AND avoid common failure modes
with other CNS equipments) (4) support a variety
of datalink services (incl. addressed
Air-Ground, addressed Air-Air and broadcast) (5)
support different ground infrastructures (support
a variable air-traffic density and connectivity
of ground stations)
5
Provide a sustainable migration path (1)
Mode S extended squitter and STDMA for ADS-B
E-TDMA for AOC, ATSC ADS-B and ASAS-C
VDL Mode 2 for AOC and ATSC
ACARS for AOC and some ATSC
2000 2005 2010 2015 2020
1
channels possibly available for VDL in Europe
2 or More ?
6

• limit the cost of aircraft equipments
• adopt a GSM-like cellular deployment model, to
  emit
• a single datalink channel at any location in
  airspace
• support safety-oriented certification
• rely on a built-in Statistical
  Self-Synchronisation
• design a stable cellular layout after
  constraints set by air traffic density, and the
  required throughput and transfer delays
• provide QoS management mechanisms at every
  service interface
• design the medium access control policy to
  guarantee if necessary the transit delay
  performance for time-bounded ATSC transactions
  (ADS, CPDLC, ASAS...)

7

• cells tailored to operations
• air traffic density
• deployed applications
• en-route, TMA, airport
• cellular layout description
• loaded as pre-flight information
• periodically broadcast on a GSC
• handover protocol
• aircraft-initiated (based on the cellular layout
  and own position)
• inter-connected ground stations
• self-insertion mechanism
• for popping-up aircraft
• as a backup or alternative to handover

8
Comparison with VDL Mode 3 4 ()
Percentage of a channel required to meet the QoS
requirements
Solution 3
Solution 1
Solution 2
(European Scenarios)
TMA Short-Term
52.6 (40.1 )
54.1 (41.6)
32.1 (26.1)
TMA Medium-Term
176.6 (120.4)
181.3 (125)
66.7 (54.6)
66.4
En-Route Short-Term
69.3
51.3
En-Route Medium-Term
306.1 (287.4)
315.5 (296.8)
301.8 (239.1)
160 NM radius, 570 aircrafts max, 95 satisfied
transfer delay, D8PSK modulation Solution 1
represents the ground-centralised VDL Mode 3T
(data-only mode) Solution 2 represents a modified
VDL Mode 4 (10 seconds frame instead of 1
mn) Solution 3 represents a variation of E-TDMA
without guaranteed access delay, parenthesised
figures were obtained after downgrading a
certain QoS category
() source Dassault Electronique with Alcatel
Bell, National Avionics and IAA Final
Report of the TREATY 8 study funded by CEC DG 13
(delivered in april 98)
9
Provide a sustainable migration path (2)
VHF band for aeronautical communications (25 kHz
channels)
data channels
today
Protected sub-band for 8.33 kHz voice channels
2005
25 kHz channels freed by the migration into the
sub-band
2010
Protected E-TDMA clusters (to minimise the
interference problems channels used by adjacent
cells should belong to different clusters)
10
Hybrid ATM concept combining global
(re-)planning and local autonomy
air-air data exchange (ADS-B ASAS)
local semi-autonomous conflict-solving
air-ground ATN links for ATC and tactical
re-planning
discrete rendezvous points defining a 4D Flight
Contract
11
The foreseen E-TDMA Traffic Mix
required downlink throughput
required uplink throughput
mobile-originated emissions ATSC
AOC ADS-B emissions from
ASAS the ground station(s)
ATSC AOC
12
Limit the aircraft equipment cost
the autonomous aircraft equippage upgrade for an
air-air ASAS/ADS-B capability consists of 2
additional receivers tuned on downstream cells
forward listening to adjacent cells
talking and listening in the current cell
13
Summary description of some features proposed
in the E-TDMA study
14
Statistical Self-Synchronisation (S3)

• a robust, low-cost synchronisation is a crucial
  issue
• UTC accuracy for applications must be 1 s
• D8PSK VDR must not drift by more than 50 µs
• E-TDMA solution
• no constraint on the VDR (just a quartz clock)
• no external master clock (ground or space-based)
• global coordination among all stations not
  needed
• can use imprecise position information (RNP
  level)
• low overhead (a few percents of the capacity)
• confined to a distinctive synchronisation
  sublayer

15
Statistical Self-Synchronisation (S3)
Oh I'm late, I should speed up
Fine, I can slow down a bit

• Each station detects if it is "late" or "early"
  by monitoring the emissions of the other
  ones (coarse position information can be used
  to improve the correlation of delays)
• It shifts its time back or forth by a small
  quantum when certain guard time thresholds are
  no longer respected
• Some extra guard time is needed for the
  resynchronisation

16
High integrity MAC sublayer

• expected Physical Bit Error Rate 10-3
• existence of error bursts (due to fading)
• required Residual Message Error Rate 10-7
• limit cost of real-time error processing
• E-TDMA solutions
• interleaving for scattering error bursts
• a small number of combinable BCH and RS modules
• target Undetected Error Rate 10-5 to 10-6
• additional CRC at LLC layer with target RER lt
  10-7

17
support end-to-end safety certification
include a strong error detection/correction
within the MAC sublayer to minimise losses of
end-to-end integrity and repetition delays
ERROR gtTRANSPORT NACK AND RETRANSMISSION
ERROR gtLL NACK AND RETRANSMISSION
IS
IS
ES
ES
18
performance certification in an ATN context
offer an ISO 8208 service interface QoS
params (low overhead in local reference mode for
ATN) support QoS selection parameters at the
serviceinterface of the E-TDMA subnetwork
(allowing the ATN routers to establish SVCs
according to QoS)
SVC2 (QoS3)
ground router
aircraft router
SVC1 (QoS2)
SVC3 (QoS1)
19
Modular error correction
an E-TDMA slot would combine only a few different
codes defined according to the target Residual
Error Rates (RER)
header blocks B0 RER 3.10-8
BCH (31, 16)
small slots B1 RER 3.10-6
BCH (63, 45)
large slots B2 RER 10-6
RS (31, 23) 5 bits symbols
example for a small slot B0 3B1 gt 151 data
bits 69 CRC bits CRC overhead 45 (worst
case based on BER 10-3) example for a larger
slot B0 9B2 gt 1051 data bits 335 CRC
bits CRC overhead 30 (worst case based on BER
10-3)
20
support different datalink services
ATN IP
(sub)network Layer
LL sub-layer
broadcast
MAC sub-layer
SYNCH sub-layer
Physical Layer
21
QoS monitoring

• there is a need to monitor and report all the
  problems so as to allow
• the E-TDMA system to self-reconfigure quickly
  whenever necessary
• (switching a whole cell to a backup frequency
  when the current one
• becomes too disturbed to be relied on for safe
  ATM operation)
• E-TDMA solution
• the mobile stations broadcast event reports on
  any serious incident
• like the non reception of emissions from the
  ground station(s)
• the ground station(s) manage counters and
  averagers to determine
• the duration (number of cycles), the extension
  (number of users)
• and the intensity (number of slots) of the
  perturbation, according
• to its own monitoring activity and the reports
  sent by the mobiles
• the ground station(s) and/or the mobiles use
  decision thresholds in a sliding observation
  window to trigger a change of frequency

22
Backup frequency switching protocol
E-TDMA solution

• the cellular frequency plan (including backup
  frequencies)
• is a priori known by everybody (eg broadcast on
  the GSC)
• warm restart the ground station uses the uplink
  part of the cycle
• to broadcast on the new frequency the list of
  all the mobiles, that
• confirm their presence in their primary slot,
  and so in one cycle a
• normal situation is re-established (normal
  case)
• cold restart if the situation seems abnormal
  (e.g. collisions occur
• on primary slots) the ground station(s)
  invite(s) the mobiles to use
• the insertion-and-echoback protocol (with more
  Hello mini-slots
• offered than in the standard situation) in
  order to come back to a
• coherent state in a few cycles

23
The E-TDMA frame (1)
frame (N-1) frame (N) frame
(N1)
E-TDMA cycle
E-TDMA cycle
E-TDMA cycle

• the frame duration must not be larger than
• either the ADS-B period, or
• the minimax access time to be 100
  guaranteed
• E-TDMA cycle between 2 and 10 seconds
  (depending on local requirements)

24
The E-TDMA frame (2)
individual slot structure
next slot
CRC and decay
propagation guard time (3.3 µs / km S3 guard)
ramp-up and synchro (1.9 ms)
data
total slot duration
25
The E-TDMA frame (3)
QoS1
QoS2
QoS0
exclusive primary slot for ADS-B and short urgent
messages
shared secondary slots for other messages
(longer and less frequent ones)
26
The E-TDMA frame (4)
QoS0
QoS1
QoS2
SYNCH
SYNCH
uplink slots can be left contiguous and the QoS
breakdown remains virtual (dynamically managed by
the ground station) intermediate
resynchronisation beacons may be
needed (depending on the drift performance of VDR
clocks) the initial guard time may be halved
(since the ground station is at the center of the
cell)
27
E-TDMA QoS categories
1) exclusive primary slot QoS0
mean transit time E-TDMA cycle / 2 max transit
time E-TDMA cycle period E-TDMA cycle min
throughput slot length / E-TDMA cycle
2) shared secondary slot QoSi , i 1, ... n
Ki slots shared between N aircraft, with an Li
average percentage load
mean transit time (Li / 100) (N / 2Ki)
E-TDMA cycle max transit time (N / Ki) E-TDMA
cycle min throughput slot length / (N E-TDMA
cycle / Ki) available throughput 100 / Li min
throughput period (N / Ki) E-TDMA cycle
28
MAC protocol (1)
E-TDMA deterministic slot assignment scheme

• every secondary slot of QoSi is shared between
  all the aircraft
• that have the same primary slot number modulo
  Ki, Ki being the
• maximum number of secondary slots available for
  QoSi.
• when the maximum number of aircraft N is
  reached, at most
• N/Ki aircraft may queue up for each slot

7 4 1
... 5 2
... 6 3
1 2 3 4 5 6 7 ...
3 shared secondary slots

• relaxing the modulo Ki constraint yields less
  deterministic solutions which are still
  completely collision-free owing to this
  distributed queueing system

29
MAC protocol (2)
E-TDMA distributed per-QoS-scheduling solution

• reservation flags are set in the primary slots
• the ground station echoes-back the reservations
• the reservation order rules are implicit yet
  unambiguous

b
reservation echo-back
a b
a
shared pool of secondary slots
reservation flags in the primary slots
30
Handover versus Self-Insertion (1)
fully coordinated ground infrastructure
en route cells (continuous tessellated coverage)
airport-centered cell
31
E-TDMA air-initiated ground-coordinated handover

• the aircraft knows approximately her position
  (published RNP) and
• the cellular structure and she initiates the
  handover automatically
• if the RNP capability is lost, the handover must
  be initiated manually

I'll now switch to station B
Hello, this is A/C x
A/C x
1
5
on B, you have slot N, 'bye
A/C x comes in on slot N
4
3
give me a slot for A/C x
2
station A
station B
32
Handover versus Self-insertion (2)
loosely coordinated ground infrastructure
en route cells (continuous coverage)
airport-centered cell
boundary
33
Self-insertion protocol (1)
successful insertions are echoed-back by the
ground station(s)
p-persistent CSMA access scheme on a small set of
Hello mini-slots to be used by candidate
aircraft not handed over by another station
p-persistent CSMA/CD and acknowledgment module
34
Handover versus Self-insertion (3)
no ground infrastructure
low density cells without ground stations
low density multi-station macro-cell
coast line
35
The autonomous mode must adapt the design
principles of the E-TDMA concept
to operational situations when no ground
infrastructure exists the air traffic density
is low the E-TDMA is used only in the
air-air local mode (broadcast or addressed)
36
E-TDMA in the autonomous mode
Statistical Self-Synchronisation (S3) Fixed
E-TDMA cycle in each cell Fixed cellular
layout Support different datalink services High
integrity datalink Deterministic MAC
sublayer Self-insertion mechanism Distributed QoS
monitoring
features that nedd to be adapted to the
autonomous mode
37
The autonomous E-TDMA Traffic Mix
required "downlink" throughput
ADS-B ASAS-B ASAS-C other air-air services (eg
SIGMET-B) Network and QoS Management services
38
The distributed round robin scheme

• every secondary slot of QoSi is shared between
  all the aircraft
• that have the same primary slot number modulo
  Ki, Ki being
• the maximum number of secondary slots
  available for QoSi
• when the maximum number of aircraft N is
  reached, at most
• N/Ki aircraft may queue up for a slot

7 4 1
... 5 2
... 6 3
3 shared secondary slots
1 2 3 4 5 6 7 ...
39
Self-insertion in autonomous mode
Incoming mobiles broadcast arrival messages into
free primary slots (no dedicated insertion
slots) Arrival messages are re-emitted by
other mobiles across the whole cell (echo-back
or handovers by means of ground stations are not
available) A rebroadcast counter is decremented,
to stop the propagation process after a finite
number of hops Collisions between the
cycle-simultaneous arrivals are also
(re)broadcast by the other mobiles, with
a collision rebroadcast counter, allowing to sort
out the collisions between non-simultaneous
arrivals
40
Distributed QoS Monitoring
The primary slots carry additional System
Management information a slot occupancy
bitmap (as consolidated by the mobile)
fields for broadcasting short messages that
describe some special anomalous events, as
detected by the receiver part unexpected loss
of air-air connectivity with another mobile,
erroneous data transmission, severe signal
jamming... every mobile monitors its
environment, and it may trigger QoS alarms
(broadcast to the other mobiles, and also sent
to the cockpit) when some threshold is crossed
(eg error rate)
41
Propagation scheme for self-insertion
the arrival is notified 5 cycles later at the
other end of the cell
the incoming aircraft broadcasts an
arrival message in a number of free slots
C4
C0
C3
C2
C1
autonomous E-TDMA cell requiring 5 propagation
hops
42
Back-propagation of collisions (1)
incoming aircraft loses the slot
incoming aircraft loses the slot
CC2
CC2
CC 0
CC1
CC1
C4
CC2
CC0
CC1
C'4
C3
C'3
C2
C'2
shortest path for collision rebroadcast
43
Back-propagation of collisions (2)
the earliest incoming aircraft retains the
slot (the collision rebroadcast does not reach
her)
the latest incoming aircraft loses the slot
CC0
CC1
C4
CC0
CC1
CC0
CC1
C'4
C3
C2
C1
C'3
shortest path for the collision rebroadcast