Leakage Rate (LR): Metrics & Trade-Offs

• Leakage Rate is a metric that measures the transfer of physical quantities (information, mass, charge, or population) across boundaries with domain-specific definitions and formulas.
• It underpins analyses in diverse fields such as wireless security, quantum key distribution, semiconductor device performance, and hydrodynamics by guiding system design and trade-offs.
• Leakage Rate introduces crucial trade-offs between throughput and security, influencing error correction, privacy protocols, and practical measurement methodologies in engineering.

Leakage Rate (LR) quantifies the transfer of physical quantities—information, mass, charge, or population—across boundaries that are ideally impermeable or confidential. The exact metric, operational meaning, and mathematical representation of Leakage Rate are domain-specific but universally serve as critical performance constraints in security, sensing, engineering, and quantum information. This article presents a rigorous cross-disciplinary synthesis of LR, delineating its definitions, analytical forms, and tradeoffs in communication security, quantum key distribution, device physics, quantum error correction, privacy-constrained information theory, and classical hydrodynamics.

1. Fundamental Definitions and Domain-Specific Metrics

The term leakage rate appears in a range of precise forms across scientific domains:

• Information-Theoretic Leakage Rate (Wireless Security): In quasi-static fading wiretap channels, the average information Leakage Rate $R_L$ is the expected number of confidential bits per channel use revealed to an eavesdropper. For Rayleigh-fading single-antenna wiretap channels under fixed-rate wiretap coding (code rate $R_b$, secrecy rate $R_s$), it is given by:

$$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$

where $\mathrm{Ei}(\cdot)$ is the exponential integral, $\bar{\gamma}_e$ is Eve's mean SNR (Huang et al., 2018).

• Quantum Key Distribution (QKD): The Leakage Rate $LR$ is the classical information per raw-key bit about Alice's sifted key revealed to an adversary due to information reconciliation. The raw definition is

$LR = \frac{\mathrm{leak}}{t}$

where $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$ and $C$ is the reconciliation message. In Slepian-Wolf-optimal protocols, $R_b$0 (Elkouss et al., 2013).

• Classical Device Leakage Rate (Physical Systems): In hydrodynamics, leakage rate $R_b$1 ($R_b$2) is the volumetric flow rate through a boundary (e.g., pipeline leak, seal, fractured formation):

$R_b$3

For seal microchannels, $R_b$4 (Poiseuille), $R_b$5 is critical separation, $R_b$6 fluid viscosity (Lorenz et al., 2010).

• Semiconductor Device Leakage Rate: For irradiated silicon diodes and LGADs, leakage rate is the reverse dark current due to SRH bulk generation and trap-assisted transport:

$R_b$7

(Yang et al., 2021).

• Quantum Information Leakage Rate: In quantum processors, the LR is the probability per operation that population leaves the computational subspace, e.g., for a CPTP noise channel $R_b$8 on $R_b$9:

$R_s$0

(Wu et al., 2023, Xin et al., 21 Nov 2025).

• Privacy and Source Coding: For communication schemes with privacy constraints, the leakage rate generalizes to the normalized mutual information or maximum guessing probability:

$R_s$1

(Goldfeld et al., 2015, Yakimenka et al., 2021).

2. Analytical Approximations and Computation

Leakage Rate formulas often involve intractable integrals or optimization over code and channel parameters; tractable closed forms are critical:

• Approximate Average Leakage Rate in Fading Channels: For $R_s$2, $R_s$3, the main exponential-integral difference can be approximated for large $R_s$4 as $R_s$5, yielding

$R_s$6

enabling analytic transmission design (Huang et al., 2018).

• QKD Reconciliation Leakage: Asymptotically, for LDPC-based rate-adaptive protocols, $R_s$7 with $R_s$8 the effective code rate; $R_s$9 is the Slepian–Wolf limit (Elkouss et al., 2013).
• Seal Microfluidics: For rough seals, leak-rate $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$0 is governed by the critical-junction theory, $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$1, with $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$2 determined self-consistently via surface roughness spectra and contact mechanics (Lorenz et al., 2010).
• Quantum Gate Leakage: Benchmarked via leakage randomized benchmarking protocols, experimental survival probabilities decay exponentially as $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$3 with exponents related to $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$4 (Wu et al., 2023, Xin et al., 21 Nov 2025).

3. Trade-Offs Involving Leakage Rate

Leakage Rate is an explicit constraint or penalty, introducing fundamental performance trade-offs:

• Throughput–Leakage Trade-off (Wiretap): For fixed-rate wiretap codes, maximizing throughput $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$5 under $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$6 yields

$$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$7

Throughput saturates as the leakage constraint is relaxed; excessive security margin does not improve data rate (Huang et al., 2018).

• Privacy–Rate–Distortion–Leakage (Source Coding): With download rate $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$8, distortion $$R_L = \frac{1}{\ln 2}e^{1/\bar{\gamma}_e}\left[\mathrm{Ei}\left(-\frac{2^{R_b}}{\bar{\gamma}_e}\right) - \mathrm{Ei}\left(-\frac{2^{R_b-R_s}}{\bar{\gamma}_e}\right)\right],$$9, and maximal leak $\mathrm{Ei}(\cdot)$0, the tradeoff obeys

$\mathrm{Ei}(\cdot)$1

with boundary cases

$\mathrm{Ei}(\cdot)$2

(Yakimenka et al., 2021).

• Broadcast Channel Capacity–Leakage Region: Leak constraints $\mathrm{Ei}(\cdot)$3 enter Marton-like achievable rate regions by modifying the private-message bounds:

$\mathrm{Ei}(\cdot)$4

and similarly for $\mathrm{Ei}(\cdot)$5; at $\mathrm{Ei}(\cdot)$6 the region reduces to standard confidentiality, at large $\mathrm{Ei}(\cdot)$7 to the unconstrained case (Goldfeld et al., 2015).

• Quantum Error Correction: Leakage rates directly constrain logical error suppression in code families; subsystem codes achieve considerably better logical protection per qubit under high leakage than subspace surface codes and retain effective distance up to threshold leakage rates $\mathrm{Ei}(\cdot)$8 (Brown et al., 2019).

4. Experimental Methodologies for Leakage Rate Measurement

Application-appropriate measurement protocols are domain-defined:

Domain	Measurement Protocol	Quantitative Output
Wireless Security	Analytical/numerical evaluation over SNR statistics	$\mathrm{Ei}(\cdot)$9 in bits/channel use
QKD	Classical message-length, smooth min-entropy chain rule	$\bar{\gamma}_e$0 per raw key bit
Semiconductor Devices	TCAD-SIMS calibrated current-voltage characteristics	$\bar{\gamma}_e$1 (A)
Quantum Gates	Leakage (Interleaved) Randomized Benchmarking (LRB/iLRB)	$\bar{\gamma}_e$2, $\bar{\gamma}_e$3
PMT Enclosures	Accumulation-box SF$\bar{\gamma}_e$4 trace with sniffer, calibration runs	LR in Pa·m$\bar{\gamma}_e$5/s (SF$\bar{\gamma}_e$6/He equiv)
Fluidic Seals, Pipelines	Volumetric tracking (pressure drop, flow rate, transient analysis)	$\bar{\gamma}_e$7 (m$\bar{\gamma}_e$8/s)

For large-scale PMT electronics, a polycarbonate accumulation-box with SF$\bar{\gamma}_e$9 tracer and ppm sniffer provides a detection sensitivity of $LR$0 Pa·m$LR$1/s (SF$LR$2), well below the requirement for water ingress prevention (Chu et al., 30 May 2025).

5. Leakage Rate in Quantum Information and Device Physics

• Gate-Induced Leakage in Quantum Processors: For superconducting transmon processors, the LR per entangling gate (e.g., CZ, iSWAP) is extracted from conditional-oscillation experiments or leakage RB:

$LR$3

Embedding a "Leakage Removal Unit" (LRU)—concurrent dispersively-tuned pulses—achieves removal fractions $LR$4 with negligible impact on computational-readout fidelity. Maintaining low leakage rates is essential to suppressing logical errors in memory and stability benchmarks (Xin et al., 21 Nov 2025).

• Passive Leakage Removal and Transport: Disordered transmon arrays support passive leakage evacuation with an engineered balance between leakage propagation ($LR$5) and last-site reset by feedback measurement or dissipation. There are two optimal measurement rates: $LR$6 (propagation-limited) and $LR$7 (disintegration-limited). Quantitatively, steady-state leakage population decays as $LR$8, so $LR$9 (Martín-Vázquez et al., 20 Feb 2025).
• Code Performance Under Leakage: Subsystem surface codes (Bravyi–Bacon–Shor type) confine leakage-induced correlated errors spatially. For swap-based leakage reduction (swap-LR), subsystem codes outperform subspace codes below threshold rates $LR = \frac{\mathrm{leak}}{t}$0 for DP-leakage and $LR = \frac{\mathrm{leak}}{t}$1 for MS-leakage when using Bacon–Shor codes (Brown et al., 2019).

6. Device and Engineering Applications

• Semiconductor Detectors: LR modeling for LGADs under high neutron fluence incorporates both local acceptor removal and global midgap trapping, with post-irradiation scaling

$LR = \frac{\mathrm{leak}}{t}$2

where $LR = \frac{\mathrm{leak}}{t}$3, robustly matching measured currents at $LR = \frac{\mathrm{leak}}{t}$4C to within 5–10% (Yang et al., 2021).

• Pipeline and Fluid Transport: In water/gas pipelines, LR is tied to orifice outflows. For unsteady gas dynamics in parallel pipes, the leak rate is modeled as a time-decaying function:

$LR = \frac{\mathrm{leak}}{t}$5

where $LR = \frac{\mathrm{leak}}{t}$6 is the measured inlet pressure, yielding better correspondence to transients than static leak models (Aliyev et al., 11 Sep 2025). Extended Kalman Filter assimilation of transient pressures using a 24D hydraulic state vector enables accurate (<2% error) estimation of leak positions and rates in noisy environments with minimal sensors (P et al., 2021).

• Drilling and Well Integrity: Time-dependent LR during lost-circulation events in fractured formations is derived from Herschel–Bulkley rheology, giving

$LR = \frac{\mathrm{leak}}{t}$7

where $LR = \frac{\mathrm{leak}}{t}$8 solves a nonlinear ODE involving fracture geometry and fluid parameters. Dimensional curve-sets ("type curves") allow fast field diagnosis, and Monte Carlo methods yield probabilistic LR/confidence bands for real-time operations (Albattat et al., 2020).

7. Leakage Rate Constraints in Information Theory and Security

• Broadcast Channels with Leakage Constraints: LR empirically links physical-layer security and capacity: a leakage constraint $LR = \frac{\mathrm{leak}}{t}$9 introduces a penalty term and additive relaxation in achievable rates. Inner bounds based on Marton coding and outer TV-approximation lemmas precisely delineate permissible tradeoffs. As $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$0, one recovers secrecy-capacity; for $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$1 large, standard Marton/UVW bounds are restored (Goldfeld et al., 2015).
• Private Information Retrieval: In loss-tolerant PIR, the optimal download rate $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$2 trades off permissible average leakage $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$3 (ML guess probability), reconstruction distortion $\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$4, and communication per file. With exact privacy ($\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$5) the rate matches that of downloading all files; with no privacy constraint ($\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$6) one recovers classical source coding (Yakimenka et al., 2021).

---

Summary Table: Primary Leakage Rate Metrics and Operation Domains

Context	LR Definition and Metric	Operational Implication
Wiretap channel	$\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$7: bits/use, average equivocation loss	Security-throughput optimization
QKD reconciliation	$\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$8: leak per raw bit, min-entropy reduction	Limits on secret key length
Quantum gate/circuit	$\mathrm{leak} = H_0(C) - H_\infty(C|X^{(t)})$9: pop. leakage per gate/measure	Logical error rates, QEC thresholds
Semiconductor device	$C$0: reverse current density	Radiation tolerance, device lifetime
Hydrodynamics/devices	$C$1: volumetric flow ($C$2/s)	Integrity/safety monitoring
Privacy constraints	$C$3, $C$4: MI or guess probability	Download cost, information-theoretic security
Broadcast channel	$C$5: normalized MI ($C$6)	Achievable rates under secrecy budget

---

Leakage Rate is thus a central, domain-specific parameter that encapsulates leakage in various physical, information-theoretic, and engineering contexts, dictating operational security, code design, device integrity, privacy guarantees, and information flow. Its analytical expressions, measurement protocols, and optimization tradeoffs serve as foundational tools for the design and analysis of secure, robust, and high-integrity systems across science and engineering.