AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


WPA password cracking: Parallel Processing on the Cell BE

Author

Term

10. term

Publication year

2009

Abstract

Dette projekt undersøger, hvordan man implementerer et WPA-PSK password-crackingværktøj på PlayStation 3 med Cell Broadband Engine (CBE). Først gennemgås WPA (Wi‑Fi Protected Access) standarderne for at identificere deres potentielle svagheder. Analysen peger på, at WPA-PSK (pre-shared key) har én velkendt angrebsmulighed under selve godkendelsesprocessen. For at bygge en prototype blev relevante dele af den åbne kildekode fra wpa_supplicant udtrukket og samlet til et fungerende program på en PC. Profilering af koden viste, hvor parallel behandling kunne udnyttes, og hvilke dele der var mest beregningstunge. Herefter blev værktøjet portet til PlayStation 3: først til Power Processing Unit (PPU) med små justeringer for håndtering af Big-endian datalagring, dernæst optimeret til Synergistic Processing Units (SPU'er). Ydelsen blev benchmarked på PC, PPU, én SPU og SPU'er efter optimering. En klynge af seks SPU'er leverede næsten 30 gange flere passwordforsøg pr. sekund end en implementering kun på PPU. Efter en anden implementeringsrunde, der inddrog SHA1-SSE2 kildekode, nåede klyngen af seks SPU'er op på 701 testede passfraser pr. sekund. Arbejdet viser, hvordan PS3'ens parallelle arkitektur kan accelerere beregningstunge opgaver som password-afprøvning.

This project explores implementing a WPA-PSK password-cracking tool on the PlayStation 3’s Cell Broadband Engine (CBE). It begins by reviewing WPA (Wi‑Fi Protected Access) standards to identify potential weaknesses. The analysis indicates that WPA-PSK (pre-shared key) has a single well-known attack surface during the authentication process. To build a prototype, relevant components from the open-source wpa_supplicant were extracted and combined into a working program on a PC. Code profiling revealed opportunities for parallel processing and pinpointed the most compute-intensive sections. The tool was then ported to the PlayStation 3: first to the Power Processing Unit (PPU) with minor adjustments for big-endian data representation, and later optimized for the Synergistic Processing Units (SPUs). Performance was benchmarked on the PC, PPU, a single SPU, and SPUs after optimization. A cluster of six SPUs achieved nearly 30 times more password attempts per second than the PPU-only implementation. After a second iteration incorporating SHA1-SSE2 source code, the six-SPU cluster reached 701 tested passphrases per second. The work demonstrates how the PS3’s parallel architecture can accelerate compute‑heavy tasks such as password testing.

[This abstract was generated with the help of AI]