Byte parity kept all six-bit SEEG decisions in eight offline records under one known packet loss and correct received bytes

We tested byte parity for a fixed SEEG reader in eight public records from OpenNeuro ds004473 version 1.0.1. SEEG means stereoelectroencephalography. Seven contact difference payloads and one XOR parity payload used the same packet size as eight ring differences. Early trials set the shared scales. Later trials supplied 1,128 baseline and response windows. At six bits per real or imaginary value, parity kept every compressed score and decision in 9,024 repeated single-loss tests. The receiver knew the lost packet, and all received bytes were correct. The ring changed 503 decisions. Both methods used 664 bytes per window, including headers. This was 78.61 percent below a native float size reference with the same header budget. Compression itself changed 64 native decisions. Mean balanced accuracy was 63.50 percent for parity and 63.62 percent for native input. The ring had 63.92 percent mean single-loss accuracy. Thus the supported benefit is stable output after loss, not higher mean task accuracy. A later check showed that the sum of rounded ring differences determined its contact change after loss. Fresh source ranges matched one retained window per person. A full repeat analysis matched all main results. The package contains inputs, models, packed packets, scores, source records, code, three figures, and repeat instructions. XOR parity is an established code. No live control or brain stimulation was tested.

stereoelectroencephalography · packet loss · XOR parity · signal compression · brain-computer interfaces · error correction