Optimality of Stationary and Deterministic Policies for Zero-Delay Coding of ℝ^d-Valued Linear Markov Sources with Quadratic Cost

Zero-Delay Lossy Coding of Linear Vector Markov Sources: Optimality of Stationary Codes and Near Optimality of Finite Memory Codes

“…Two important structural results for the finite horizon problem (1) have been developed by Witsenhausen [3], and Walrand and Varaiya [4]. These results are stated in the following two theorems.…”

“…Email: 18mg16@queensu.ca, tamas.linder@queensu.ca, yuksel@queensu.ca. This work was presented in part at the 2020 IEEE Conference on Decision and Control [1] at each time t ≥ 0, as noted in [2], the coding policy selects a quantizer Q t : R d → M based on past information (q [0,t−1] , X [0,t−1] ), and then quantizes X t as q t = Q t (X t ). Because of this, we refer to Π as a quantization policy.…”