Doug Cooper
Canada Country Manager
Intel Corporation
March 20, 2007
Mr. Cooper
My name is Andre Brisson and I am a founder of Whitenoise Laboratories Inc.
I want you to consider another “product line” and to do so I need to use a couple of examples, each
brought up with the utmost respect.
First, Whitenoise is the fastest encryption ever but it is truly characterized by the fact that it provides
continuous stateful authentication of persons and devices, inherent detection, and automatic denial of
network access to hackers and thieves.
I propose that Intel should “dust off” one of its older chips and utilize Whitenoise to create a SCADA
specific chip.
Analogy –
Moore’s law at some point will become obsolete simply because there will become a saturation point
when humans are not generating data at exponential rates any more.
Until then, it seems as if it is an expensive proposition to follow the Gillette model of product releases:
one blade, then two, then three, then four and now five blades to a razor! For chips it becomes daunting
to fathom the progression from chip, to dual cores, quad cores etc. etc. Margins become progressively
harder to maintain as the processors get more sophisticated and expensive.
I have attached an article from a couple of years ago when Texas Instruments surpassed Intel for the
first time in total chips sold. They dusted off an old chip and sold it to Nokia.
SCADA, Systems Control and Data Acquisition, networks absolutely dwarf in size what we consider
to be the “human networks”. A small SCADA network will have 150,000 endpoints. Machine to
machine communications control all of our critical infrastructure.
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
1
Because of the characteristics of Whitenoise, particularly its negligible processing overhead, its
incredible speed, and offset management, this security technology can be deployed in an archaic Intel
chip and easily outperform anything in the market today. INTEL could market a line of chips
particularly aimed at the SCADA markets. The cost to INTEL to rollout would be a fraction of the cost
of embarking on a next generation chip.
We can provide verification of all metrics and we have world class security experts you can reference.
http://www.wnlabs.com/advisors.html
There is much to learn about Whitenoise at http://www.wnlabs.com/technology.html.
I would like you to consider the potential of a paradigm that we have developed called Dynamic
Distributed Key Infrastructure. On the CD provided are several presentations. One has chip
characteristics and was a presentation used recently with another chip manufacturer.
California is looking towards utility companies monitoring energy consumption in homes so they don’t
have the rolling blackouts they recently ran into etc.
Their crypto needs for this project are listed below. Whitenoise can be used just for authentication, or
the same key can also be used for encryption, detection, authorization, DRM, RNG etc. As you read
the crypto requirements for this SCADA proposal, consider that Whitenoise satisfies all of their
requirements without doing a hybrid of symmetric keys with asymmetric key update (PKI). If you read
the attachment called SFI2_AES_DIVA.pdf you will see that we can accomplish the same thing but
with the Whitenoise process of distributed keys in turn distributing more session keys. This
tremendously simplifies the architecture, performance costs etc.
Thank you for any time you spend looking into Whitenoise technologies and a possible collaborative
effort on making chips for SCADA devices. This system can be demonstrated for you via SFI2 by
Lofty Perch (www.loftyperch.com ). And it eliminates problems of relying on time stamps etc.
Sincerely,
Andre Brisson CEO
www.whitenoiselabs.com
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
2
SCADA space example
Specs for utility project http://sharepoint.californiademandresponse.org/pct/default.aspx
Issues for Cryptographic Measures
The working group has identified the following issues concerning cryptographic measures:
The consensus of the group is that encryption of the broadcast
Confidentiality
vs. Authentication messages is not necessary for confidentiality purposes; only
authentication is necessary.
While encryption is not considered necessary for purposes of
Authentication
confidentiality, it has been suggested that it may require less
Methods
processing power to encrypt the entire message as a form of
authentication than to use another method of authentication and
integrity checking, e.g. a cryptographic hash.
Because the thermostat may pass through many hands: utility,
Symmetric vs.
manufacturer, installer, customer, etc., it seems likely that
Asymmetric
asymmetric keys would be very useful in this application.
Cryptography
However, the amount of processing required is orders of magnitude
higher, and the memory and code footprint required could raise
costs considerably.
Is has been suggested that there be multiple levels of keys, with the
Periodic Key
lowest level being symmetric and periodically changed using a
Changing
higher level key to avoid eavesdropping attacks. Considering how
infrequent the event messages are, however, it is not clear how
often the keys would need to be changed.
It has been suggested that it would be preferable to have multiple
Number and
Grouping of Keys sets of keys so that the number of customers affected by the
compromise of any particular key or set of keys would be limited.
It has been suggested that key sets be randomly distributed among
thermostats sold so that any attack could not be concentrated in a
geographic area. Naturally, the more key groups there are, the
more difficult it will be to manage the keys.
An appropriate mechanism for recovering from the compromise of
Recovery from
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
3
Key Compromise
one or more keys is necessary. The method must not involve travel
to the customer site. So far, the methods considered include using
multiple levels of keys as discussed above, or using a finite number
of keys preloaded in the thermostat.
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
4
Cryptographic Approaches Considered
The task force has considered the following general categories of cryptographic security solutions.
The first two fall in the category of “traditional cryptography”. Details of the actual solution will vary,
but this description should give the main ideas of each approach.
Option 1: Symmetric Session Keys with Asymmetric Update
This approach follows current best practices without consideration of the amount of processing
power or memory footprint required by asymmetric cryptography calculations.
a. Authentication and integrity are provided via an HMAC on each message. Alternatively,
this could be done by encrypting the entire message.
b. The HMAC is calculated using a symmetric Session Key and appropriately changing data
to protect against replay.
c. The Session Key is periodically changed and the new value broadcast at intervals,
encrypted using an Update Key.
d. The Update Key is asymmetric.
e. The Public Update Key is encoded in the thermostat when it is manufactured. This will
avoid concerns about the security of the Update Key in transit between the sender and the
manufacturer.
f. The Private Update Key is held by the sender.
g. There may be multiple sets of Update Keys, with the set used by any thermostat chosen
randomly when it is manufactured. To affect all thermostats, the sender must transmit
multiple messages.
h. There may be another level of asymmetric keys, the System Keys, used to change Update
Keys if they become compromised.
Option 2: Symmetric Key Combination
This approach uses only symmetric operations and would be based on methods commonly used
in military applications.
a. The manufacturer chooses a random number and codes it into the thermostat before
shipping. Call this value “A”.
b. The manufacturer also encodes a well-known fixed value “D”.
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
5
c. The manufacturer supplies a set of serial numbers and corresponding A values to the
sender.
d. The sender chooses a second random value for that thermostat. Call this value “B”. There
could be a different B value for each thermostat, or the thermostats could be grouped so that
many of them have the same “B”.
e. The installer phones the sender and provides a serial number.
f. The sender provides the installer with a “B” value to be entered into the thermostat.
g. The sender combines A and B to produce C. Any number of functions could be used to
perform the combination, the simplest being an XOR.
h. The sender authenticates each message using an HMAC (or similar mechanism) of the
message concatenated with D, using C as a key.
i. The sender must transmit a different message for each A/B/D combination. Grouping can
be achieved by varying any of the three values.
j. A possible variation would be to “force” B to achieve the same C for various groups. This
would potentially reduce the number of messages that must be broadcast.
Option 3: TESLA or Variation
Research has indicated that authentication in broadcast networks can be achieved with low
processing requirements using one-way chains of symmetric keys and periodic, secure time
synchronization (“Timed Efficient Stream Loss-tolerant Authentication” or TESLA) A
method using some variation of this scheme would be the third option.
Andre Brisson
Whitenoise Laboratories Inc.
Executive Vice President
Business Development (Co-founder)
Phone: 604-724-5094 Fax: 604-873-2467
Email: brisson@lightspeed.ca
6