In this work, we design an interactive coding scheme that converts any two party interactive protocol Π into another interactive protocol Π , such that even if errors are introduced during the execution of Π , the parties are able to determine what the outcome of running Π would be in an error-free setting.Importantly, our scheme preserves the space complexity of the protocol, in addition to the communication and computational complexities. Specifically, if the protocol Π has communication complexity T , computational complexity t, and space complexity s, the resulting protocol Π is resilient to a constant > 0 fraction of adversarial errors, and has communication complexity approaching T as approaches 0, computational complexity poly(t), and space complexity O(s log T ).Prior to this work, all known interactive coding schemes required the parties to use at least Ω(T ) space, as the parties were required to remember the transcript of the conversation thus far, or considered weaker error models.
Introduction
Interactive Coding TheoryBackground. In the well-studied field of coding theory, which dates back to the seminal work of Shannon [24], researchers attempt to understand the fundamental limits on the transfer of information imposed by unreliable communication channels. Most work in this regime focused on the one-way model of communication, where one party (henceforth referred to as Alice) wishes to send a single message to a second party (Bob). While classical coding schemes have found innumerable applications both in theory and in practice, modern systems, which are often very interactive, have motivated the development of radically new coding schemes.Motivated by this scenario, Schulman [23] initiated the study of the following model of interactive communication over a noisy channel. There are two parties who wish to carry out a conversation. The additional wrinkle: the channel through which the parties communicate is now unreliable, and may change some of the sent symbols. Therefore, the goal is to transform the original protocol into a new protocol which is still guaranteed to correctly determine the outcome of the original protocol (or, say, succeed with high probability), even if some errors are introduced into the conversation. The new protocol is referred to as a robust simulation of the original protocol. In the literature, errors may be random or adversarial, and in our work, we consider the most general adversarial error model, so our results can be applied in all other (weaker) error models.Resource-efficient interactive coding. Schulman's breakthrough works in the 1990's [21,22,23] already showed that every protocol can be robustly simulated by a protocol that only incurs a constant multiplicative overhead in the communication complexity, even in the case that an adversary is allowed to corrupt a constant fraction of the total communication. It would be another twenty years before Brakerski and Kalai [1] show that the robust simulation can also be made computationally efficient. That is, the running...