Constant-Rate Interactive Coding Is Impossible, Even in Constant-Degree Networks

Multiparty interactive coding allows a network of <inline-formula> <tex-math notation="LaTeX">$n$ </tex-math></inline-formula> parties to perform distributed computations when the communication channels suffer from noise. Previous results (Rajagopalan and Schulman, STOC 1994) obtained a multiparty interactive coding protocol, resilient to random noise, with a blowup of <inline-formula> <tex-math notation="LaTeX">$O(\log (\Delta +1))$ </tex-math></inline-formula> for networks whose topology has a maximal degree <inline-formula> <tex-math notation="LaTeX">$\Delta $ </tex-math></inline-formula>. Vitally, the communication model in their work forces all the parties to send one message at every round of the protocol, even if they have nothing to send. We re-examine the question of multiparty interactive coding, lifting the requirement that forces all the parties to communicate at each and every round. We use the recently developed information-theoretic machinery of Braverman <italic>et al.</italic> (J. ACM 2018) to show that if the network’s topology is a cycle, then there is a specific cycle task for which any coding scheme has a communication blowup of <inline-formula> <tex-math notation="LaTeX">$\Omega (\log n)$ </tex-math></inline-formula>. This is quite surprising since the cycle has a maximal degree of <inline-formula> <tex-math notation="LaTeX">$\Delta =2$ </tex-math></inline-formula>, implying a coding with a <italic>constant blowup</italic> when all parties are forced to speak at all rounds. We complement our lower bound with a matching coding scheme for the cycle task that has a communication blowup of <inline-formula> <tex-math notation="LaTeX">$\Theta (\log n)$ </tex-math></inline-formula>. This makes our lower bound for the cycle task tight.

[1]  Rafail Ostrovsky,et al.  Optimal Coding for Streaming Authentication and Interactive Communication , 2015, IEEE Transactions on Information Theory.

[2]  N. S. Barnett,et al.  Private communication , 1969 .

[3]  Guy Kindler,et al.  Lower Bounds for the Noisy Broadcast Problem , 2008, SIAM J. Comput..

[4]  Mark Braverman,et al.  Constant-Rate Coding for Multiparty Interactive Communication Is Impossible , 2017, J. ACM.

[5]  Leonard J. Schulman,et al.  A coding theorem for distributed computation , 1994, STOC '94.

[6]  Noga Alon,et al.  Reliable communication over highly connected noisy networks , 2016, Distributed Computing.

[7]  Yael Tauman Kalai,et al.  Interactive Coding for Multiparty Protocols , 2015, ITCS.

[8]  Alexander A. Sherstov,et al.  Optimal Interactive Coding for Insertions, Deletions, and Substitutions , 2017, 2017 IEEE 58th Annual Symposium on Foundations of Computer Science (FOCS).

[9]  Mark Braverman,et al.  Coding for Interactive Communication Correcting Insertions and Deletions , 2017, IEEE Transactions on Information Theory.

[10]  Leonard J. Schulman,et al.  Communication on noisy channels: a coding theorem for computation , 1992, Proceedings., 33rd Annual Symposium on Foundations of Computer Science.

[11]  Bernhard Haeupler,et al.  Interactive Channel Capacity Revisited , 2014, 2014 IEEE 55th Annual Symposium on Foundations of Computer Science.

[12]  Amit Sahai,et al.  Efficient Coding for Interactive Communication , 2014, IEEE Transactions on Information Theory.

[13]  Leonard J. Schulman Coding for interactive communication , 1996, IEEE Trans. Inf. Theory.

[14]  Leonard J. Schulman,et al.  The Adversarial Noise Threshold for Distributed Protocols , 2014, SODA.

[15]  Ran Gelles,et al.  Coding for Interactive Communication: A Survey , 2017, Found. Trends Theor. Comput. Sci..

[16]  Ran Gelles,et al.  Making Asynchronous Distributed Computations Robust to Channel Noise , 2018, ITCS.

[17]  Ran Raz,et al.  Interactive channel capacity , 2013, STOC '13.

[18]  Amit Sahai,et al.  Adaptive protocols for interactive communication , 2013, 2016 IEEE International Symposium on Information Theory (ISIT).

[19]  Robert G. Gallager,et al.  Finding parity in a simple broadcast network , 1988, IEEE Trans. Inf. Theory.

[20]  Bernhard Haeupler,et al.  Optimal Error Rates for Interactive Coding II: Efficiency and List Decoding , 2013, 2014 IEEE 55th Annual Symposium on Foundations of Computer Science.

[21]  Madhu Sudan,et al.  Optimal error rates for interactive coding I: adaptivity and other settings , 2013, STOC.