When we think about transmitting information, we usually focus on the content of the message—the bits packed into a data packet. However, a powerful alternative channel exists entirely in the temporal dimension: when a message is sent can be just as informative as what it contains.
By establishing a shared rule between a sender and a receiver, the mere occurrence of an event at a specific moment can transmit complex state information implicitly, requiring few to no explicit bits. Yet, this implicit communication channel possesses a fundamental vulnerability: it is highly sensitive to delay. When latency enters the system, the temporal information degrades, triggering sharp phase transitions where the advantage of timing information completely vanishes.
The Intuition of the Silent Channel
Consider a sensor monitoring an unstable system. In a traditional periodic setup, the sensor transmits the exact state of the system at rigid time intervals (e.g., every millisecond). This requires a continuous stream of explicit data bits.
Now imagine an event-triggered setup. The sensor remains completely silent until the system's state drifts and hits a predefined boundary threshold. The moment the boundary is breached, the sensor fires a single pulse.
Even if that pulse contains zero payload bits, the receiver learns two critical pieces of information the moment it arrives:
- The system state has reached the exact magnitude of the threshold.
- It reached that threshold exactly now.
As synthesized by Michael Hertneck et al. in Current trends and future directions in event-based control, this mechanism allows for system stabilization at significantly lower explicit data rates than periodic sampling. By treating the temporal occurrence of a transmission as an implicit communication channel, systems can drastically save bandwidth and energy. This principle extends beyond hardware sensors; for instance, John Williamson et al. demonstrate in Designing for uncertain, asymmetric control: Interaction design for brain-computer interfaces that the precise timing of discrete events can serve as a robust method for encoding human intent even when the underlying physical channel (like an EEG) is incredibly slow and noisy.
How Delay Acts as Temporal Noise
If timing information is so efficient, why do we ever use periodic sampling? The answer lies in how communication latency interacts with time-encoded data.
When a channel introduces an unknown delay, it acts exactly like noise on a traditional signal, but instead of distorting the amplitude, it distorts the clock. If a sensor triggers an event at time $t$, but the message does not arrive until $t + \Delta$, the receiver no longer knows the exact moment the threshold was crossed. The implicit information begins to decay.
In The value of timing information in event-triggered control: The scalar case, Mohammad Javad Khojasteh et al. uncover a fascinating phase transition governed by this latency. When communication delay is minimal, the implicit timing channel is highly accurate, allowing a controller to stabilize an unstable plant with a near-zero explicit bit rate. However, as the delay increases, the timing information becomes "stale." To compensate for the growing uncertainty of when the event happened, the sensor is forced to start packing more explicit data bits into its payloads.
Low Delay -------------> High Timing Information Value ----> Near-Zero Explicit Bits Needed
High Delay -------------> Stale Timing Information ---------> High Explicit Bits Needed
The Critical Threshold and Phase Transitions
This degradation is not a gradual, linear decline. Instead, networked control systems exhibit a hard mathematical boundary known as the critical delay threshold.
As proven by Mohammad Javad Khojasteh et al. in Time-triggering versus event-triggering control over communication channels, this threshold represents the exact point where the information gained from the timing of an event-triggered message is completely wiped out by the uncertainty of the network latency. Beyond this critical value, the efficiency advantage of event-triggered communication completely evaporates, and the system performs no better than traditional periodic control.
When operating under significant latency, the controller must over-communicate explicit data bits to counteract the stale timing information, as detailed in The Value of Timing Information in Event-Triggered Control and Exploiting Timing Information in Event-Triggered Stabilization of Linear Systems With Disturbances. A similar system-level boundary is identified by Chi Harold Liu et al. in Toward QoI and Energy-Efficiency in Internet-of-Things Sensory Environments, where the relationship between signaling latency and task urgency dictates a hard boundary for system feasibility; if the time required to coordinate sensors exceeds the application's delay tolerance, the quality of information collapses entirely.
Parallel Phase Transitions in Wireless Channels
Remarkably, this non-linear threshold behavior of delayed information is not unique to event-triggered control loops; it mirrors a fundamental phenomenon in wireless information theory.
In Multi-Input Single-Output (MISO) broadcast channels, a transmitter relies on Channel State Information (CSIT) to optimize data rates to multiple users. When this feedback is delayed, its utility undergoes a strikingly similar phase transition based on its prevalence.
As established by Ravi Tandon et al. in On the Synergistic Benefits of Alternating CSIT for the MISO Broadcast Channel and Two-user MISO broadcast channel: Synergistic benefits of alternating CSIT, strategically alternating between different states of channel knowledge creates a "synergistic gain." However, the operational value of this delayed feedback is highly non-linear:
- When delayed information is scarce: Its marginal utility is incredibly high, matching the potency of perfect, instantaneous channel state information.
- When delayed information is abundant: A threshold is crossed where it becomes entirely obsolete, making the channel behave as if the transmitter had no state information at all.
This exact threshold behavior carries over directly into physical layer security. Pritam Mukherjee et al. demonstrate this in MISO broadcast channels with confidential messages and alternating CSIT and Secrecy for MISO broadcast channels via alternating CSIT, proving that the secure data rates achievable via alternating CSIT are strictly bound by these same non-linear abundance thresholds.
Conclusion
Whether stabilizing a physical plant over a wireless network or optimizing multi-user communication channels, time itself functions as a high-value currency. Leveraging the implicit information encoded in when an event occurs allows systems to bypass traditional data-rate bottlenecks. However, because time flows relentlessly forward, any delay in transmission acts as a destructive force, creating sharp phase transitions that dictate exactly when timing information is a priceless asset—and when it becomes completely worthless.
Key Foundational Work
On the Synergistic Benefits of Alternating CSIT for the MISO Broadcast Channel (2013) — contributed 1 concept elements to this direction.
Designing for uncertain, asymmetric control: Interaction design for brain-computer interfaces (2009) — contributed 1 concept elements to this direction.
Two-user MISO broadcast channel: Synergistic benefits of alternating CSIT (2013) — contributed 1 concept elements to this direction.