Re: encryption algorithm - Algorithms Discussion Boards

and it won't be just lemmings going over it if he ever gives me the wrong formula to work with again...... (perhaps I should have known better... D'Oh! | :doh:

)

"More functions should disregard input values and just return 12. It would make life easier." - comment posted on WTF

"This time yesterday, I still had 24 hours to meet the deadline I've just missed today."

First rule of debugging: don't trust anything or anyone.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

this months tips:
- use PRE tags to preserve formatting when showing multi-line code snippets
- before you ask a question here, search CodeProject, then Google

Luc Pattyn wrote:
First rule of debugging: don't trust anything or anyone

Good point Big Grin | :-D

"Find it your bloody self - immediately!" - Dave Kreskowiak

at least the debugger! Especially if there is a method in the Watch window that has side effects D'Oh! | :doh:

-^-^-^-^-^-
no risk no funk ................... please vote ------>

As a general policy, members on this site do not respond to questions that are obviously homework. You should also assume that your instructor visits this site and has read your post. If he hands out unique assignments, you may need to change your user name to DeadMeat.

Hello all;

I'm currently working on encryption algorithm for small embedded systems.....I had selected a combination of RSA and Blowfish.....where RSA is used for the session key encryption...

Could any body suggest a method of transferring the public keys to the recipients with authentication without using the CA certification or PKI....as its dealing with embedded systems and having a third party to monitor the public key transfer is a major area of concern..

Any help in this regard will be greatly appreciated..

Thanking you'll in advance..

sc_emb

sc_emb

Be your own CA and install your public master key as trusted on the embedded system. Then the embedded system can trust all certificates signed by your master private key. Much the same as IE - it comes with a whole lot of certificates pre-installed from CAs that Microsoft has decided that you should trust.

If you don't use assymetric encryption for authentication you have to save some secret key in each device that could be vulnerable to detection.

Of course it all depends on what you are securing, if the embedded system is a nuclear detonation device or if failure of the system could otherwise lead to mass destruction, then this advice may not be applicable.

Peter
"Until the invention of the computer, the machine gun was the device that enabled humans to make the most mistakes in the smallest amount of time."

Perhaps I’m just going over semantics here but...are not public keys supposed to be public? The public key for your device can be public. Your device needs the public keys of authorized senders. And perhaps its own RSA key pair if it needs to say much in return.

In the following E(key, msg) is RSA encryption of msg with public key and D(key, msg) is RSA decryption using public key. E’() and D’() are the same except they use the private RSA key.

msg = D’(key, E(key, msg))

Alice: (plain text) I am Alice, an authorized sender.
Device: Oh? Prove it. Here is a time stamp...
Alice sends D’(keyAlice, Oct 9 2007 2228)
Note – Alice uses her private DEcryption key here.
(Device computes E(keyAlice, above message. All is okay.)
Note – Device encrypts to decrypt. I know it sounds strange.
Alice sends E(deviceKey, long random Blowfish key) + Blowfish encrypted message using the just sent key.
Device recovers Blowfish key with D’(deviceKey, key part of prior message)

The point is simple. The device needs only the public keys of trusted senders. If you need a CA for that then you need a CA.

You might find that Twofish is easier to implement. In particular you do not need over 4000 bytes of pi. And a place to keep them for the next encryption. But if you have a hard drive that should not be a problem. On a machine large enough to be reasonable to do Twofish I would hardcode q0 & q1 and generate a table of already permuted MDS stuff on the fly. I.e. 512 bytes of hard coded constants + 4096 bytes of table generated as part of your key setup.

Twofish ain’t that complex if I can implement the thing in both x86 and ARM assembly. I’ll do it for the Z80 or a PIC with several hundred bytes of RAM if someone will pay me. I bought Schneier’s book “used.” It made that nice cracking sound when I first opened it. My overall comments on Twofish: it appears more than a little over designed. And you don’t need 3 different encode / decode procedures for different key lengths.

. . . .

Oh wait. After writing the above I just reread your original message. Apparently you have multiple devices needing public keys.

You can only allow trusted public keys into your system. Start with your public key. Your machines trust you. Only you or someone you delegate may enter new authorized users. In effect YOU are the CA.

Don’t forget about physical access and how that can compromise your devices.

-Peter

i have a mathematical function,example x^2+sin(x),but it is String.i want it is mathematical function,it can return value float

leo

nta_3886 wrote:
i want it is mathematical function,it can return value float

Ok.

"A good athlete is the result of a good and worthy opponent." - David Crow

"To have a respect for ourselves guides our morals; to have deference for others governs our manners." - Laurence Sterne

use a stack to convert the formula into Reverse Polish Notation (RPN). Then evaluate the RPN version. You can find many examples of how to do both steps of this via google.

--
If you view money as inherently evil, I view it as my duty to assist in making you more virtuous.

dan neely wrote:
use a stack to convert the formula into Reverse Polish Notation (RPN). Then evaluate the RPN version. You can find many examples of how to do both steps of this via google.

Semi-on-topic... I just found my code for this from Pascal class back in the 80's. I used it on my first job to build an RPN based spreadsheet program in RPG II. Smile | :)

_________________________
Asu no koto o ieba, tenjo de nezumi ga warau.
Talk about things of tomorrow and the mice in the ceiling laugh. (Japanese Proverb)

a lot of work D'Oh! | :doh:

Russell

If you are using the .net Framework then you can compile that into a method in a temporary assembly and execute it as normal code.

I'm sure you can find a sample about that in the WWW.

-^-^-^-^-^-
no risk no funk ................... please vote ------>

Hi,

if your expression uses a .NET language such as C# you can have .NET compile and run it
for you, as in http://www.codeproject.com/csharp/evalcscode.asp[^]

if you want to define your own langauge you will need to parse and evaluate everything
yourself, as in http://www.codeproject.com/csharp/sota_expression_evaluator.asp[^] or http://www.codeproject.com/cpp/MathieuMathParser.asp[^]

Did you notice all links are to existing CodeProject articles, which you can search
using the Search box available on nearly every CP page?

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

this weeks tips:
- make Visual display line numbers: Tools/Options/TextEditor/...
- show exceptions with ToString() to see all information
- before you ask a question here, search CodeProject, then Google

If you are using Rexx then

interpret say "string"

Except you first need to write sin(x) and other functions you use in Rexx. You also need to use Rexx rules for syntax.

Rexx can compile and execute on the fly.

With just a little more detail in dos one might type

rexx math 1.2*4.7

5.64 prints

The file math.cmd contains

/* execute incomming statement */
numeric digits 40
arg cmd
interpret say cmd

It's been a while but saving x to a environment variable or some such is not that hard.

My 40 digit ln() in ln.cmd is

/* ln(x) */
numeric digits 40
arg x

/* prescale to put x into reasonable range */

xp = 0
if x > 2 then do until x < 2
xp = xp + 1
x = x / 2
end

/* now that x <=2 compute ln(x)
method from Abramowitz etal section 4.1.39
ln(2) to 40 places takes 26 loops */

z = x-1
val = 0
do t = 26 to 1 by -1
t2z = t*t*z
val = t2z / (t+t+val+1)
val = t2z / (t+t+val)
end
val = z / (val+1)

/* account for prescale */

if xp = 0 then return val
return (const(ln2)*xp) + val

Hello,

Suppose that we have a NxN square filled with letters. Which algorithm is the best to find any words starting with NiNj and continue from the starting point through adjacent squares (left-right-cross) where the word is in any length L?

Thanks and best regards..

.:: Something is Wrong ::.

I wrote a decent 5x5 Boggle game and felt that the problem was the dictionary. So...I kept multiple partial matches going always working the partial match with the lowest lexical value.

Finds game buried in some oddball directory... Discovers it does not work in xp. Posting an old result...

A L E R E
O O T S T
T G L U T
N Z O Y E
S C E N F

alert
alertest
alertly
ales
aloe
alter
alto
alts
colog
cols
colt
colter
coly
cone
coney
cony
coulter
couter
cozen
cozey
enfetter
enol
erst
ester
eyen
eyne
fets
fetter
fetus
feus
fytte
glue
gluey
glut
gluten
goal
goer
goes
gone
gout
gouty
lest
lets
logo
loot
looter
lots
lotus
lust
luster
lustre
lusty
lute
nets
nett
netter
neuter
nolo
nous
oleo
oles
oots
ousel
oust
ouster
outer
outgo
outgoes
outre
outs
outsert
outset
oyez
reset
rest
result
retool
rets
retuse
scene
scone
scouse
scout
scouter
sere
sloe
slog
slot
slue
slut
stela
steno
stere
stereo
stey
stoa
stole
stool
strette
stutter
stye
suet
suety
sutler
tela
terse
test
tester
testy
toes
tola
tole
tolu
toluene
tool
tooler
toot
tooter
tootle
tootler
tote
toter
tots
trestle
tret
tuts
tutty
tyne
ulster
user
utter
yett
youse
zone

So ?

.:: Something is Wrong ::.

pashje wrote:
Which algorithm is the best

It depends on the value of N and on the lenght of the dictionary (L)...

If N is too big (compared with L) the best way is to try to search every word of the dictionary on the NxN square.
Elseway (N small) try to extract every possible words, and then test if the word exist.

BTW I prefere in both cases to use the first method. It seemes easyer to be implemented. Big Grin | :-D

Russell

Well both methods will do the job surely but what if N is a Z+ ? It may 3 but also 3000.. So the performance is an important issue..

Maybe a genetic algorithm or somemthing with trees will work well ? Any text about thiss issue will really help !;)

.:: Something is Wrong ::.

if N can change from 3 to 3000...well...I think that it is not possible to find an algorithm with always an High performance. You need to implement at least 2 algorithms: the program will use the first one if N<k and="" the="" other="" if="" n="">K.
The treshold K must be set during your test to find what algorythm is faster.

About genetic algorithms I think that this isn't the right way.
I think that you can perform a tree algorithm sorting the dictionary somehow to avoid to check every words.

Russell

I did my final year project on a scrabble game and i found that the best way to increase speed when matching words is by choosing an efficient data storage technique for your dictionary.

A tree structure is needed to solve this problem, a DAWG would probably be the best choice.

Hi everyone,

Could anyone kindly assist with some code (preferably Java) for the Travelling Salesman Problem that uses any of the following Heuristics
1. Tabu Search
2. Hill Climbing
3. GRASP (Greedy Randomized Adaptive Search Procedure)
4. Simulated Annealing

I need this to just text some problems.

Thanks

laremtj

Jesus Loves You, Give your Heart to Him Today

laremtj wrote:
preferably Java

This is a Visual Studio/.NET Centric site, if in case you haven't noticed.

"Any sort of work in VB6 is bound to provide several WTF moments." - Christian Graus