Pehlivanoğlu, Meltem KurtSakallı, Fatma BüyüksaraçoğluSakallı, Muharrem Tolga2024-06-122024-06-1220212147-0030https://search.trdizin.gov.tr/yayin/detay/511853https://hdl.handle.net/20.500.14551/13721Abstract—In this paper, we generate new hardware efficient involutory 32 × 32 binary Maximum Distance Separable (MDS) diffusion layers with branch number 5. In our construction method, the idea used in Generalised Hadamard (GHadamard) matrix form is applied when generating these diffusion layers. We construct lightweight circuits by applying Boyar’s global optimization heuristic (BP) to these diffusion layers. Hence, new 32 × 32 binary involutory MDS matrices with the best-known implementation cost (78 XORs) and depth 4 are generated. The obtained result is the same with the previous result given in [1], and we show that the diffusion layer given in [1] can also be obtained directly by using our construction method. As a result, we give thirteen more new involutory 32 × 32 binary MDS matrices with the best-known resulteninfo:eu-repo/semantics/openAccessOn the Construction of Low-latency 32 × 32 Binary MDS Matrices from GHadamard MatricesArticle104111118511853