Cryptography Against Continual Memory Leakage

Recall from last lecture that we have several ways to model leakage. One model is “only computation leaks” by Micali and Reyzin [11], which assumes a form of secure memory that does not leak as long as no computation is done on the data. Another one is “memory leakage” by Akavia, Goldwasser, and Vaikuntanathan [1], which assumes that everything can leak information. From an orthogonal dimension we can talk about bounded leakage and continual leakage, where the former assumes that leakage is just one shot, while the latter allows leakage to happen repeatedly during computation. The table below summarizes all four possible combinations of leakage models. only computation leaks memory leakage