MAC

The planes are blind. The only way they can 'see' one another is to hear each other's broadcast positions.

Radio Bumpers
This concept has been advanced to avoid collisions between planes. Low power, very brief ping packets are broadcast by each plane. The signal strength is extremely low - if you can hear the transmission, you are too close!

Timing
The very nature of this concept is an extremely low power signal - detectable only in immediate proximity. Interference is a small problem in this case. In fact, there is a sense that each plane could be constantly broadcasting its position. However. a plane that is broadcasting cannot receive. Although some form of scheduled or slotted transmission seems attractive, they quickly become unwieldy. Simple random timing seems to be the most reliable way to get messages through.

Using GPS, each plane can determine its absolute 3D location. Multiple GPS readings quickly yield the plane's heading and speed.

This position and vector are the content of the minimal IMHERE message. This content is wrapped in a header that includes the broadcaster's identification and timestamp.

A more sophisticated implementation might include other elements, such as Kalman filter terms that specify the multidimensional uncertainty of the data.

A receiving plane which is also tracking its own GPS coordinates, can determine proximity - form the position data. From the vector, it can determine whether the two planes are on a collision course.

No reliance on GPS
But this approach does not rely on GPS. In battle conditions, GPS systems will be attacked, spoofed and jammed. General GPS failure - or local failure in either plane - will disable the precision system illustrated above.

The simplest approach is to simply replace the information supplied by GPS with the same coordinates, derived from inertial systems. Conceptually this is transparent to the collision avoidance system and should permit the system to ignore GPS outages.

The problem is that the inertial system alone is unlikely to be accurate enough to permit close formation flight for a long period of time.

Alternatively, very precise, but limited, data can be determined simply by comparing the transmit and receive timestamps on the message. This will not yield heading or even relative position. It merely determines the proximity of the broadcasting plane. It can be quiockly seen if a plane has penetrated the radius of safety.

Comparison of multiple distance readings will show determine how quickly the two planes are approaching one another (or diverging) . It will not be able to determine - even if planes are rapidly approaching - whether they are on a collision course.

NINS

A broader, and very attractive, solution to the problem of GPS failure is to immediately erect a networked inertial navigation system.

In this model, each plane has imprecise navigational information from its own inertial guidance system. However, it can accurately measure distances to other planes, based on signal timestamps. With multiple interlocking radii, a tight map can be generated in which all the planes have accurately known positions in a common local coordinate system.

This accurate local system will be immediately useful to resolve the collision avoidance problem being considered here.

More generally it can be used to create an accurate navigational system by pooling the information of all the aircraft. A filter that sums all the inertial systems and removes error will be far more accurate than any one of its constituents. The common inertial guidance that emerges from this system would permit better (and far better coordinated) navigation than an individual INS.

It is important to point out that Networked INS is a different concept than Radio Bumpers. Radio Bumpers is based on very low power broadcasts, detectable only by planes approaching the safe flying radius. On the other hand NINS requires high-power transmission, reaching the maximum number of aircraft.

In an extreme situation there may not even be a shared clock to use for distance measurements. For this case, we might consider an interesting (but crude) fallback technigue.

The receiving plane can use simple signal strength to estimate distance. Since it diminishes with the square of distance it would give reasonably accurate estimates even when using a system hse sensitivity is likely to be logarithmic.

This fallback technique requires to changes to the system. The easy change is to the IMHERE packet, which does not now include a measure of signal strength at the transmitter. Since this is variable under software control, the broadcasting plane would need to include this information in the packet.

A more significant change is to the communication hardware - or at least to the API with which the Communication system interacts with the Command and Control unit. Currently the signal strength with which a packet is received is not reported. (It may not even be measured.) Only the level of error-correction required by the packet is reported.

While a system feature like signal strength reporting might not be justified by this application alone, there are several capabilities that would be enabled (or enhanced) by a metric of signal strength (and/or signal to noise ratio). We hope to make a persuasive argument for this feature in the near future.

IMHERE	ID	TIMESTAMP	POSITION			VECTOR			ERRCORR