20.500.12556/DKUM-93352 Dvofazna optimizacija daljših binarnih zaporedij z nizkimi avtokorelacijami magistrsko delo Dual-Step optimization for long binary sequences with low autocorrelation Problem iskanja binarnih zaporedij z nizko avtokorelacijo (angl. low autocorrelation binary sequences problem) predstavlja izjemen računski izziv, saj je klasificiran kot težek kombinatorični problem. Binarna zaporedja z visokimi merit faktorji, in s tem nizkimi avtokorelacijskimi lastnostmi, imajo pomembne aplikacije v digitalnih komunikacijah, kjer omogočajo učinkovito ločevanje signalov od šuma, pa tudi v fiziki, kemiji, kriptografiji, itd. V zaključnem delu bomo predstavili nov stohastični dvofazni algoritem za optimizacijo daljših binarnih zaporedij z nizkimi avtokorelacijami. Prva faza predstavlja paralelni algoritem, ki izkorišča popačeno simetrijo in razrede omejitev ter uporablja grafične procesne enote za pohitritev računanja. Druga faza pa je algoritem s prioritetno vrsto, ki dodatno izboljša zaporedja prve faze s tem, da sprosti omejitvi in deluje nad celotnim iskalnim prostorom problema. Pokazali bomo tudi, da dvofazna optimizacija omogoča iskanje boljših binarnih zaporedij, zlasti za daljše dolžine zaporedij. The problem of finding low autocorrelation binary sequences presents a significant computational challenge, as it is classified as a hard combinatorial problem. Binary sequences with high merit factors, and thus low autocorrelation properties, have important applications in digital communications, where they enable effective separation of signals from noise, as well as in physics, chemistry, cryptography, and other fields. In this work, we will present a new stochastic dual-step optimization algorithm for long binary sequences with low autocorrelation. The first step consists of a parallel algorithm that exploits skew-symmetry and restriction classes, utilizing graphics processing units to accelerate computations. The second step is a priority queue-based algorithm, which further improves the sequences from the first step by relaxing both constraints and operating over the entire search space of the problem. We will also show that the dual-step optimization finds better binary sequences than those currently known, particularly for longer sequences. binarna zaporedja avtokorelacija samoizogibni sprehod merit faktor binary sequences autocorrelation self-avoiding walk merit factor true true false [B. Pšeničnik] Slovenski jezik Angleški jezik Magistrsko delo/naloga 2025-06-20 12:16:44 2025-08-13 14:28:45 2025-08-19 03:19:17 0000-00-00 00:00:00 2025 0 Maribor 0 Maribor 1 spletni vir (1 datoteka PDF (IX, 35 str.)) 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.421.5(043.2) 245852931 MAG_Psenicnik_Blaz_2025.pdf MAG_Psenicnik_Blaz_2025.pdf 1 8976B09EF872923632CCB302A161B00D e685e157904a2d2af5987a34a431c3ed9b56b439b15613d94ef917ff0e4ce5af 4d1b8f11-4dc0-11f0-a0a9-00155d000105 https://dk.um.si/Dokument.php?lang=slv&id=192524 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0