Code-Based Public Key Encryption Decoding Algorithm Patents

Code-based public key encryption, exemplified by the McEliece and Niederreiter constructions, hides a message by encoding it with an error-correcting code and deliberately injecting an error pattern that only the holder of the private key can remove. Decoding — recovering the original message by correcting that injected error, typically via syndrome decoding — is the computationally hard step that gives the scheme its security, so most of the claimable invention sits in how that decoding step is structured, accelerated, or combined with other functions. Patent claims in this field tend to describe either the decoding algorithm itself or a system that uses it inside a larger communication or verification flow. Understanding which of those two claim styles dominates the prior art is essential before drafting a new application.

The assignee ranking covers 36 companies and research institutions across the 65 records in scope, with the leading assignee holding 7 records. Filing is concentrated rather than evenly spread: the top 5 assignees together account for 43.1% of all 65 records, and the top 10 account for 69.2%. That leaves a long tail of single- or few-filing entrants, which is typical of a field still anchored by a small number of specialist and academic filers rather than broad industry adoption.

Filings rose from 4 in 2007 to a peak of 12 in 2022, then fell to 1 by 2024 — a 75% decline over that 2021-to-2024 span. Because publication typically lags actual filing by around 18 months, the 2025 and 2026 figures in this dataset are still incomplete and should not be read as confirming a continued drop. The honest read is that the field cooled after its 2022 peak, but the most recent trend line has not yet settled.

H04L (digital information transmission) is by far the dominant classification, appearing on 92.3% of the 65 records in scope, which reflects that most claims wrap the decoding step inside a broader communication or transmission protocol. H03M (coding and code conversion) covers 20.0% of records and is the closest thing to a class dedicated to the decoding algorithm itself. Smaller shares appear in G06F (electric digital data processing, 18.5%), H04B, G06K, G06N, H04N and B82Y, each under 5%, marking areas with much thinner claim coverage.

The thinnest classification branches in this dataset — G06N (AI-based computing), G06K (data recognition), H04N (pictorial communication) and B82Y (nanotechnology) — each cover 3.1% or less of the 65 records, meaning very little claim density has built up there relative to the H04L and H03M core. A workable strategy is to tie a code-based decoding step to one of these adjacent functions, such as a learned model guiding error-locator search, rather than filing directly into the crowded transmission-layer claim pattern held by the leading assignees.