Admission Correlation Risk Score (ACRS)

GMI-TEL-033TelecommunicationsInfrastructure-DependentPattern O-6: Autonomous Coherence

Definition

Degree of demand correlation among admitted slices in the same resource pool, measured as the probability that correlated peak demand across independently-admitted slices will exceed the pool's total capacity.

The Measurement Gap

Measurement Gap and Decision

Decision Supported: Is the admission model accounting for the risk that independently-admitted slices may have correlated demand peaks, or is it assuming statistical independence that does not exist?

Measurement Gap:

Slice admission decisions are made individually. Each new slice request is evaluated against the resource pool's remaining capacity. If there is enough headroom for the new slice's peak demand, it is admitted. This works when slice demands are independent. But enterprise slices serving the same geographic market, the same industry vertical, or the same event patterns (sports events, commute hours, shift changes) have correlated demand peaks. When those peaks coincide, the aggregate demand can exceed pool capacity even though each individual admission appeared safe. This is the classic portfolio concentration risk from financial theory, applied to network resources. No current admission system computes the pairwise demand correlation between existing slices and evaluates the portfolio-level risk before admitting a new slice. ACRS formalizes this risk calculation using the same statistical framework that financial institutions use for correlated default risk: adapted to resource pool capacity rather than credit exposure.

Citations:

  • 3GPP TS 28.531, "Management and Orchestration of Network Slicing," Release 18: defines admission control procedures without incorporating demand correlation analysis.
  • IEEE Transactions on Network and Service Management, "Resource Management in 5G Network Slicing: A Survey," 2023. Surveys admission control approaches without addressing portfolio-level correlation risk.

Formula

Plain Language: For each resource pool, compute the pairwise demand correlation between all active slices using historical demand data. Then estimate the probability that simultaneous peak demand (where correlated slices peak together) exceeds the pool's total capacity. High ACRS means the portfolio has dangerous correlation that independent admission decisions did not account for.

Notation:

ACRS = P(Σ_i Demand_peak_i > Pool_capacity | correlation_matrix)

Estimated via:
ACRS ≈ P(X > Pool_capacity) where X ~ N(Σ μ_i, √(Σ_i Σ_j ρ_ij × σ_i × σ_j))

Variable Definitions:

  • μ_i = mean peak demand for slice i, from historical demand data (resource units)
  • σ_i = standard deviation of peak demand for slice i (resource units)
  • ρ_ij = pairwise demand correlation between slices i and j (dimensionless, -1 to 1)
  • Pool_capacity = total available resource capacity for the pool (resource units)
  • Assumes multivariate normal approximation for aggregate demand. Use Monte Carlo simulation for non-normal demand profiles.

Units:

Probability (0 to 1). 0 = no risk of correlated over-commitment. 1 = certain over-commitment at correlated peaks.

Sign Convention:

Lower is better. 0 means no portfolio-level risk from correlated demand.

Edge Cases:

  • Correlation structure changes over time (new tenants, changing demand patterns): recompute quarterly with rolling demand history.
  • Small number of slices (fewer than 10): pairwise correlation estimates will be noisy. Use longer demand history (6 or more months).
  • Some correlation is expected (all slices serve the same geographic market): the question is whether the admission model accounts for the expected correlation, not whether correlation exists.
  • Requires sufficient demand history per slice: minimum 90 days of demand data per slice.

Proposed formula, pending validation pilots.

At a Glance

Unit
Probability (0 to 1)
Trend Direction
Lower
Reporting Frequency
Quarterly
Category
Admission Risk and Slice Security Economics
Containment Collateral Impact Score (CCIS) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment