Kriptiranje komunikacije koristeći suparničko učenje evolucijskih algoritama

undergraduate thesis

• Title (english): Encryption of Communication Using Evolutionary Algorithms with Adversarial Learning
• Author: Ante Gazibarić
• Mentor: Domagoj Jakobović (mentor)
• Committee member: Domagoj Jakobović (predsjednik povjerenstva)
• Committee member: Marko Đurasević (član povjerenstva)
• Committee member: Karlo Knežević (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2019-07-18, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Kriptografija se bavi problemima kriptiranja i dekriptiranja informacija prilikom njihovog prijenosa nesigurnim komunikacijskim kanalom od pošiljatelja ka primatelju, kada bivaju izložene potencijalnim napadima. U ovom radu pokušalo se pronaći kriptografski algoritam pomoću optimizacije, konkretno pomoću Kartezijskog genetskog programiranja (eng. Cartesian genetic programming, CGP) koji pripada evolucijskim algoritmima. Klasična situacija u komunikaciji koja sadrži tri glavne uloge, a to su Alice ili pošiljatelj, Bob ili primatelj i Eve ili prisluškivač, prevedena je u problem optimizacije, gdje za sva rješenja postoji način određivanja njihove ocjene, tj. dobrote. Time je ostvarena mogućnost pronalaska rješenja pomoću CGP-a. CGP interno sadrži graf čvorova koji u sebi sadržavaju određene funkcije pomoću kojih vrše izmjene nad ulaznim tokom bitova. Time je ostvaren način kriptiranja, odnosno dekriptiranja. Ocjena rješenja izvedena je pomoću suparničkog učenja gdje je Alice imala zadatak kriptiranja, a Bob i Eve dekriptiranja poruka, s tim da u tom procesu Alice i Bob koriste isti ključ kojem Eve nema pristom, što ovaj algoritam čini simetričnim kriptografskim algoritmom. U takvom uređaju bolja dobrota Boba povlači bolju dobrotu Alice, a nasuprot toga, bolja dobrota Eve povlači lošiju dobrotu Alice, što upravo predstavlja glavnu karakteristiku suparničkog učenja. Kroz evoluciju CGP-a pomoću takvog način ocjenjivanja, Alice i Bob su uspjeli pronaći način komunikacije, a da pri tome Eve nije saznala koje informacije razmjenjuju, što je pokazano kroz eksperimente provedene u ovom radu.

Abstract (english)

Cryptography is concerned with encryption and decryption of information when transmitting them over an insecure communication channel from sender to recipient, when they are being exposed to potential attacks. This paper attempted to find a cryptographic algorithm using optimization, specifically by using Cartesian Genetic Programming (CGP) that belongs to evolutionary algorithms. A classic communication situation that has three main roles - Alice or the sender, Bob or the recipient and Eve or the eavesdropper, has been translated into optimization problem, where for all solutions there is a way of determining their ratings, ie fitness. This enabled the possibility of finding a solution by using CGP. The CGP internally contains the graph of nodes that contain certain functions which can change the input stream of bits. This is how encryption and decryption is done. The evaluation of the solution was done by using the adversarial learning where Alice had the task of encryption, and Bob and Eve had to decrypt the message, where Alice and Bob used the same key that Eve does not have access to, which makes this algorithm a symmetric cryptographic algorithm. In such a situation, Bob's better fitness entails Alice's better fitness, and, conversely, Eve's better fitness entails Alice's better fitness, which is precisely the main characteristic of adversarial learning. Through the evolution of CGP and by using such a method of evaluation, Alice and Bob has been able to find a way of communication such that Eve did not know what information was exchanged, as demonstrated through the experiments conducted in this paper.

• Keywords: kriptografija; optimizacija; kartezijsko genetsko programiranje; simetrični kriptografski algoritam; kriptiranje; dekriptiranje; suparničko učenje
• Keywords (english): cryptography; optimization; cartesian genetic programming; symmetric cryptographic algorithm; encryption; decryption; adversarial learning
• Language: croatian
• URN:NBN: urn:nbn:hr:168:932681
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2020-01-07 18:05:55