- this page is a stub
KG-13 was a universal digital fully-transistorised full-duplex key generator,
developed in the USA
around 1963. It was intended for the
encryption and decryption
of external generated data, such as digitised voice and facsimile
data. The KG-13 was commonly used with the HY-2 vocoder.
The KG-13 is an extremely complex and large device, measuring about
89 x 59 x 48 cm, and weighting more than 110 kg. It is comprised of 3
identical key generators, 2 synchronizers, an alarm assembly and 2
power supplies, and is based on the earlier development of
the KG-3 (transmitter) and
The key is entered into the key generators
by means of Remmington Rand computer punch cards 1 every 24 hours.
These cards are used to set the initial state of the
KOKEN shift register stages
and the tetrahedral key combining logic.
Despite its high weight, it was a huge improvent over earlier
systems, such as the KW-26, not least because of the transistor-based
circuits, which were constructed with FLYBALL modules.
The KG-13 provided Traffic Flow Security (TFS), which means that it produced
a continuous data stream, even if no message was being sent. The advantage of TFS
is that an eavesdropper will not be able to detect the beginning and end of
a message, and that the initial key is never repeated. The data that was sent
between two messages was fully random and was generated by a noisy diode,
similar to the SEND SYNCHRONIZER in the KW-7
(which did not have traffic flow security).
The KG-13 entered service around 1963 and was heavily used during the
Vietnam War, often for secure voice transmission in combination with the
HY-2 vocoder. The machine was succeeded in the 1970s
and 80s by modern smaller encyption devices, such as the KY-57
and later the KY-99. Nevertheless,
the last heavy and bulky KG-13 units were taken out of service as late as 1990.
Remington Rand computer punch cards have the same physical size
as IBM cards, but have circular punched holes, rather than IBM's
The diagram below shows and example of how the KG-13 was used with
the HY-2 vocoder. At the left is the audio
input/output, which could either be a handset or an external telephone or
audio line. The HY-2 converts the analogue voice into digital information
which is then passed on to the KG-13 for encryption. The encrypted data
is then passed to an external 2400 bps modem.
At the receiving end, the data received by the MODEM is first decrypted
in the KG-13 and then passed onto the HY-2, which then reconstructs or
synthesizes the human speech, based on the data in the 16 vocoder
channels. This results in a typical 'Donald Duck' style voice.
Although the cryptographic algorithm of the KG-13 has not yet been
published, some hints to its operating principles can be found in
publicly available and declassified NSA
material. It is known, for
example, that the crypto logic of the KG-13 is comprised from several
KOKEN shift register stages and a tetrahedral key combining logic.
The principle of the KOKEN registers is shown here:
A KOKEN register consists of several cascaded D-type flip/flops that
are clocked only when the output of the preceding state is high.
When the device is first switched on, all registers are reset so that
their outputs (Q) are set to zero. Next some of the flip/flops are
preset (Q=1), controlled by the key card. The inverted outputs
of each stage are then combined in a secret manner into a single
output bit, that is also used as the clock signal for the first stage.
The key combining logic consists of several logic AND, OR and XOR
circuits, and is configured by the current key card.
The KG-13 was one of the first fully solid-state or
transistorised devices developed
by the NSA
in the early 1960s. Like the KW-7,
it is entirely constructed
from small pre-built logic circuits, known as
FLYBALL modules, that were developed especially
for the NSA by Sylvania in Florida (US).
Each module had a distinct colour
to identify its (logic) function. The image on the right shows
some of the circuit boards of the KG-13 (one of which is missing),
each of which holds 9 pink-coloured FLYBALL modules with the actual
shift registers, or Koken registers. The yellow module contains a
single 2N404 emitter follower which is used as a clock driver
for the Koken registers.
➤ More about FLYBALL modules
Between October 1982 and January 1983, whilst working at the US Air Force,
electronics engineer James Atkinson discovered a series of serious
flaws in the KOKEN stages of the KG-13's internal pseudo random stream
generator . Atkinson had been memorizing all current and historical
circuit diagrams of the KG-13, the KY-3
and all of their FLYBALL modules, as a mental exercise.
When going over the circuit diagram in his mind,
he began to doubt its mathematical strength.
More than 20 years after the introduction of the KG-13, he was able to
prove that most of the KOKEN stages were mathematically flawed,
rendering the system compromised, and possibly leaking highly
classified information to the Russians. Although there is no direct
proof that the system had meanwhile been broken by the Soviet
Union, it has to be assumed that it was.
Embarrased as they were, the NSA
fixed it by modifying the KOKEN circuits
and issuing a Modification Work Order (MWO) for all operational and spare
machines in the field, an operation on which tens of millions of dollars
were spend .
Any links shown in red are currently unavailable.
If you like the information on this website, why not make a donation?|
© Crypto Museum. Created: Thursday 12 May 2016. Last changed: Sunday, 21 July 2019 - 20:22 CET.