Analysis of Age of Incorrect Information under Generic Transmission Delay

This paper investigates the Age of Incorrect Information (AoII) in a communication system whose channel suffers a random delay. We consider a slotted-time system where a transmitter observes a dynamic source and decides when to send updates to a remote receiver through the communication channel. The threshold policy, under which the transmitter initiates transmission only when the AoII exceeds the threshold, governs the transmitter’s decision. In this paper, we analyze and calculate the performance of the threshold policy in terms of the achieved AoII. Using the Markov chain to characterize the system evolution, the expected AoII can be obtained precisely by solving a system of linear equations whose size is finite and depends on the threshold. We also give closed-form expressions of the expected AoII under two particular thresholds. Finally, calculation results show that there are better strategies than the transmitter constantly transmitting new updates.

💡 Research Summary

The paper studies the Age of Incorrect Information (AoII) in a slotted‑time communication system where each status update experiences a random transmission delay. A dynamic source is modeled as a two‑state symmetric Markov chain with transition probability p (the analysis focuses on the case p ≤ ½, where the optimal estimator at the receiver is simply the most recently received update). The transmitter observes the source at the beginning of each slot and decides whether to send the current observation. Transmission times T are independent and identically distributed according to a generic distribution {p_t}, and the channel is error‑free. After a transmission finishes, an instantaneous ACK/NACK informs the transmitter whether the receiver’s estimate has changed, so the transmitter always knows the current estimate at the receiver.

AoII is defined as the product of a time‑penalty function f(k)=k and an information‑penalty function g(x,ĥx)=|x−ĥx|. With these choices the AoII at slot k reduces to Δ_k = k − U_k, where U_k is the most recent slot at which the receiver’s estimate was correct. Consequently Δ_k evolves as a Markov process: if the estimate is correct at slot k+1 then Δ_{k+1}=0; otherwise Δ_{k+1}=Δ_k+1. The evolution also depends on source transitions and on whether a transmission is in progress.

The authors focus on a threshold policy τ: the transmitter initiates a transmission only when the current AoII Δ_k is at least τ and the channel is idle. τ=∞ corresponds to never transmitting, τ=0 to always transmitting. Under this policy the system can be described by a discrete‑time Markov chain whose state is the triple s_k = (Δ_k, t_k, i_k). Here t_k∈{0,…,t_max−1} is the elapsed time of an ongoing transmission (t_k=0 when the channel is idle) and i_k∈{−1,0,1} indicates the channel status (idle, busy with a packet that matches the receiver’s estimate, or busy with a mismatching packet). Because the dynamics while the channel is busy are independent of the policy, the authors eliminate all “busy” states from the balance equations, reducing the analysis to the set of decision states where i_k=−1 and t_k=0. This yields a finite set of equations for the steady‑state probabilities π_Δ of the AoII value Δ.

A key technical contribution is the derivation of multi‑step transition probabilities P_{Δ,Δ′}(a), where a∈{0,1} denotes the action (no transmission or transmission). For a=0 the transition probabilities are trivial functions of p. For a=1 the probabilities depend on the random transmission time T and on the probability p(t) that the source remains in the same state for t consecutive slots; p(t) = (1−2p)^t for the symmetric two‑state chain. Lemma 1 gives closed‑form expressions for P_{Δ,Δ′}^{(t)}(1) under the assumption that every transmission succeeds within a known maximum duration t_max. Lemma 2 extends the result to a TTL‑style model where a transmission may be forced to terminate after t_max slots, possibly without success. The authors also list useful properties of these probabilities (e.g., independence from Δ when Δ′≤t_max−1 and Δ≥Δ′, shift invariance, and zero probability outside a bounded region).

With the transition matrix in hand, the stationary distribution satisfies π = π P(τ), where the action a depends on whether Δ≥τ. Solving the resulting linear system yields π_Δ for any τ. The average AoII under policy τ is then (\bar{Δ}τ = \sum{Δ=0}^{∞} π_Δ Δ). The paper provides explicit closed‑form solutions for τ=1 and τ=2, and shows that for general τ the system reduces to a finite set of linear equations of size roughly τ + t_max, which can be solved efficiently.

Numerical experiments illustrate the trade‑off. For modest source dynamics (p=0.1) and moderate transmission delays (mean ≈ 2 slots), the optimal τ lies between 1 and 3, delivering a significantly lower average AoII than either always transmitting (τ=0) or never transmitting (τ=∞). As the delay distribution becomes heavier‑tailed, larger τ values become optimal, confirming the intuition that aggressive transmission is wasteful when updates take long to reach the receiver. The results demonstrate that smart, threshold‑based transmission policies can substantially improve semantic freshness (AoII) compared with naïve strategies.

In conclusion, the paper delivers the first rigorous analysis of AoII under generic random transmission delays, establishes a tractable Markov‑chain framework, derives exact expressions for key performance metrics, and validates the superiority of threshold policies. The methodology is readily extensible to multi‑state sources, asymmetric Markov dynamics, channels with errors, or more complex cost functions, opening avenues for future research on semantic‑aware scheduling in realistic wireless and IoT networks.