The longer time required to align an inertial unit is the price to pay for its independency on any external equipment or signal, but Earth itself, and also it's larger accuracy if we consider a small period of time.
- A GPS receiver is faster, but is only one small piece in a huge system. The GPS ground segment (control and monitoring stations) and space segment (satellites) are necessary for the GPS receivers (user segment in the jargon) to work. They have multiple possibilities to fail, this critical dependency cannot be afforded by an aviation navigation system, GPS must be backed up, and this is done using inertial (can be INS, IRS, IMU or combined with air data units like ADIRU).
During inertial alignment the unit accurately senses local vertical and true north (actually Earth polar axis). The problem is that sensing true north requires knowing latitude, and determining latitude requires knowing true north. This is solved using successive approximations of both data --fortunately this iteration is convergent-- until the desired precision has been achieved.
It's slow because Earth rotation produces tiny effects at small time scale, and we haven't found more practical sensors yet. The effects depends on the distance from the rotation axis, larger at the equator than at the poles (thus alignment takes more time at high latitudes).
- GPS relies also on very widespread, weak and seemingly random signals, below the natural noise level, a situation which is usually unacceptable for reliable communication, but was chosen by US army to hide into noise. But the spread signal can actually be easily retrieved by digital signal processing as soon as the random generator algorithm and seed are known.
An IRS provides the position like a GPS receiver,
IRS also provides the horizontal plane (attitude reference) and the north direction which are inaccessible to the GPS alone. This is another reason inertial systems are still used.
IRS is much more accurate than GPS on a small period of time. INS is able to detect very quick accelerations and angular changes, well beyond the capacity of satellite navigation receivers. The quick determination of attitude parameter is actually a requirement for flight envelope protection (Airbus) and yaw damper control, to the point the rudder control is returned to mechanical linkage (removing fly-by-wire) if the three IRS fail on a Airbus aircraft.
Gimballed vs strapdown
Gimballed inertial systems principle is simple: Accelerometers are maintained oriented with Earth using a gimballed platform, and all measurements are done in the Earth reference system.
In strapdown inertial systems, accelerometers and (laser/optical) rate gyros are aligned in the aircraft reference system, and measurements are done related to this system. Except this difference, the principles for alignment are quite the same.
Overall principle of a strapdown inertial system, source
Old gimballed inertial are longer to align because gimballed gyroscopes are less precise than fixed laser gyros. For the explanations, let's assume a gimballed stable platform.
The alignment consists of getting the Earth orientation and speed, and inertial unit initial position. After alignment, current speed, attitude and heading will be determined by integrating accelerations and rotation rates and adding the result to the initial values.
Alignment time vs accuracy obtained
Alignment aims to determine two axis: Vertical of the location (direction of gravity) and true north azimuth (parallel to Earth rotation axis). For this phase, both accelerometers and gyroscopes are used. While vertical is relatively easy to determine with accelerometers, true north direction direct determination is not possible:
- True north direction cannot be sensed without knowing local latitude (local tangential acceleration), and latitude cannot be determined without knowing the direction of true north.
The practical way to proceed is: Set an initial value for latitude (it can be assumed), determine approximate north direction. Knowing approximate north, determine approximate latitude. From approximate latitude, refine north direction, etc.
Accuracy of true north bearing increases with alignment time:
Source: An Improved Alignment Method for the Strapdown Inertial Navigation System
Alignment requires the unit to be motionless, hence any erratic motion decreases inertial signal/noise ratio which slows this process (passengers boarding, wind acting on the tail surfaces, etc).
Inertial doesn't need external equipment or signal
Inertial is a stand alone device, completely independent of any external equipment or man-made signal. It relies only on Earth rotation speed and gravity. That's why it's still used for navigation in spite of existing radio aids and GPS (GPS is actually another radio aid).
Today, inertial is mostly used to backup GNSS
One may wonder why inertial systems are still used at the GPS era, in spite of their limited accuracy and long alignment procedure. The main reason is that you cannot blindly rely on GPS constellation availability. Let's list some of the possible failures.
GPS is the US solution for satellite positioning or GNSS. While defense organizations can obtain credits for different reasons, the difficulty to set up a civilian GNSS has been striking with the numerous delays encountered building Galileo, the European GNSS. It needs to design, launch and maintain the satellite constellation (more than 20 satellites are required for accuracy and availability). The constellation must be permanently monitored and controlled from ground stations all over the World.
GNSS signal can be perturbed by natural phenomena: Sun activity, weather and volcanic ashes may partially or totally block radio signal, reducing accuracy and availability. If the GNSS constellation fails, no precise GNSS receiver can work:
For standard precision use (read: limited precision), GNSS (GPS, Glonass, Galileo, Compass, Navic) are/will be compatible and depending on their location, users can switch from one system to the other, and possibly can use the associated augmentation system (Waas, Egnos, SDCM, Gagan, etc).
Precise GNSS use actually requires deciphering keys for a given constellation, which are not shared between constellation operators for obvious commercial and defense reasons.
Dependency on the US DoD and military conflicts
The GPS is a US military system. Agreements have be signed between DoD and civilian organizations for a civilian exploitation of the GPS constellation. This gives worldwide aviation some acceptable guarantees about GPS signal access.
However GPS is still vulnerable to jamming by opponents. In theory it is also sufficient to destroy the four ground control antennas to make GPS receivers useless after a few weeks.
GPS control stations
GPS is not always accurate and available.
Dilution of precision
GPS availability and accuracy is given by an indicator: dilution of precision or DOP. The larger the DOP, the less accurate the system. Accuracy is significantly degraded for a DOP larger than 3
It's worth remembering that, even without any of the problems previously listed, a maximum DOP value is not guaranteed, or even feasible, in polar regions, and is only guaranteed for 95% of the time in other regions. That was a design principle retained since the beginning, to limit the number of satellites in the space segment.
While high, 95% still means 72 minutes per day with unknown performance. Example of large DOP peak exceeding 3:
Radio signal reception
Obviously radio signals are not received underground (tunnels), underwater (submarines), and can be significantly weakened by foliage, to the point GPS is difficult to use in tall and dense forests.
When GPS is only temporary unavailable, inertial solutions can be used to maintain some position determination capability until the GPS signal is reacquired. This is very often used in car navigation systems to deal with tunnels and tall buildings. The inertial unit is a small (and very low precision) MEMS contained in the GPS receiver case or the smartphone.
Finally, let's rethink about the statement that GPS fix is nearly immediate.
That's true as long as the receiver is already aware of where the satellites are located. The constellation periodically broadcast the orbital parameters (navigation message containing almanacs, ephemeris and time corrections) for each satellite. This takes 13 minutes to get the almanacs required for a coarse positioning, at the slow rate used.
If the receiver has not yet the almanacs, or they have been cleared, then 13 minutes are required before the first fix can be done (and 13 additional minutes if the reception is interrupted while receiving the almanacs).
To get the regular accuracy, the recent ephemeris of the satellites used for the fix are required, this may take an additional few seconds. Ephemeris are updated each 30 minutes if I remember well.