Seminar: The Jacob Ziv Communication and Information Theory seminar
On the Capacity of Noisy Frequency-based Channels
Date:
March,12,2026
Start Time:
14:30 - 15:30
Location:
1061, Meyer Building
Zoom:
Zoom link
Add to:
Lecturer:
Yuval Gerzon
Research Areas:
| DNA is a promising medium for archival data storage, but in many practical settings it is difficult to synthesize and reliably sequence long molecules. This motivates storage in the short-molecule regime, where molecules are too short to carry unique indices and the pool is inherently shuffled. In such systems, a natural way to store information is through the relative frequencies, a histogram of molecule types. I will begin by introducing the frequency-based channel model and briefly reviewing the capacity picture in the idealized, noiseless identification setting. I will then move to the practically relevant noisy regime, where each sampled molecule is observed through noisy kernel, W, which introduces inter-symbol-interference (ISI), that mixes the input histogram before it is observed. The main focus of the talk is our capacity analysis for the noisy frequency-based channel, and itโs implication for the short-molecule regime. We prove a converse bound by showing that the noisy channel is a stochastic degradation of the noiseless one (via data processing). We also establish an achievable bound using Poissonization and an analysis of the resulting vector Poisson channel with mixing; the effect of noise appears as an explicit additive โpenaltyโ term that depends on the W. These results apply under explicit scaling conditions to a broad class of channels, including some of those induced by standard per-base sequencing noise models for DNA. M.Sc. student under the supervision of Asst. Prof. Nir Weinberger.
|