Bandwidth Allocation Trade-Off for Information Freshness in Integrated Sensing, Communication and Computing

Abstract

Age of Information (AoI) is recognized as a fundamental metric in networked systems to characterize the freshness of the information exchanged. For a scenario with an integrated sensing and communication pipeline, where the information to keep fresh is acquired through the assignment of bandwidth to environment sensing, a fundamental trade-off opens up: the more resources are allocated to ambient monitoring, the less channel capacity is available to communicate the sensing outcome. In this paper, we develop a comprehensive analytical model that integrates radar Cramer-Rao bounds, Nyquist sampling constraints, and pipeline stability conditions with target dynamics. We adopt an objective function that combines sensing and age-dependent position errors, and we characterize the stability region required for real-time operation. We show that bandwidth allocation is fundamentally shaped by pipeline stability rather than channel capacity alone. Further, the optimal allocation depends on the underlying target motion model, leading to distinct strategies for Wiener versus nearly constant velocity (NCV) dynamics: while linear AoI penalties favor communication-dominated regimes, cubic penalties allow operation closer to the stability boundary, allocating more bandwidth to sensing.