A reliable Turing machine

We consider computations of a Turing machine subjected to noise. In every step, the action (the new state and the new content of the observed cell, the direction of the head movement) can differ from that prescribed by the transition function with a small probability (independently of previous such events). We construct a universal 1-tape Turing machine that for a low enough (constant) noise probability performs arbitrarily large computations. For this unavoidably, the input needs to be encoded -- by a simple code depending on its size. The work uses a technique familiar from reliable cellular automata, complemented by some new ones.