COMPARISON OF SYMMETRIC CIPHERS IN C#

V diplomskem delu obravnavamo študijo simetričnih kodirnikov, ki so implementirani v knjižici Bouncy Castle in primerjamo njihovo učinkovitost kodiranja v programskem jeziku c#. V teoretičnem delu je opisana kratka zgodovina enkripcije, sledi pa ji razlaga modernih kodirnikov. Praktični del naloge obsega predstavitev testnega okolja za testiranje kodirnikov, ki smo ga implementirali v okolju .NET in analizo testiranih kodirnikov. Omejili smo se na bločne kodirnike AES, BlowFish, DES, Camellia, RC2 in Serpent ter na pretočni kodirnik HC128. Opravili smo študijo števila ciklov inicializacije, ciklov kodiranja/dekodiranja, časa kodiranja in porabe procesorja. Za vse meritve smo opravili 1000 ponovitev ter se omejili na datoteke velikosti 1 kB, 100 kB in 1000 kB. Za bločne kodirnike smo se omejili na dopolnilne načine CBC, CFB in OFB ter na 128 in 256-bitna ključa. Meritve so pokazale, da se med različnimi velikostmi datotek cikli inicializacije kodirnikov ne razlikujejo preveč, ker velikost datotek ne vpliva na inicializacijo. Cikli kodiranja pa se večajo glede na velikost datoteke. Ugotovili smo tudi, da pretočni kodirnik HC128 porabi manj procesorja v primerjavi z bločnimi kodirniki. Čas kodiranja se spreminja glede na velikost datoteke, vrste dopolnilnega načina in vrste kodirnika.In this thesis we study symmetric ciphers implemented in Bouncy Castle libraries and compare their code efficiency in the programming language c#. Theoretical part of the thesis consists of a short history of encryption, followed by an explanation of modern ciphers. Practical part of the thesis deals with introduction of testing environment implemented in .NET Framework and analysis of tested ciphers. The study was restricted to AES, BlowFish, DES, Camellia, RC2 and Serpent block ciphers, as well as HC128 stream cipher. We measured the processing time and the number of cycles spent for initialisation and encription/decryption on a standard personal computer. All measurements consisted of 1000 repetitions and were restricted to file sizes of 1 kB, 100 kB and 1000 kB. Block Ciphers were measured in CBC, CFB and OFB block cipher modes and with 128 and 256-bit keys. The results show that among different sizes of files, the cycles of initialisation of ciphers do not differ much because the file size does not influence initialisation. Cycles of coding do get bigger with bigger files. We have found also that the stream cipher HC128 uses less CPU in comparison with block ciphers. Code timing changes according to the file size, type of block cipher mode and type of cipher