Cloudflare Edge Detection

EdgeShield

Edge Protection Active
Mode: Simulation
Worker: Online
Monitoring

No active side-channel attack detected

Recent hardware counter windows are consistent with benign execution.

Low
No active attack pattern
0%

Mitigation Status

Automated edge response

🛡️

Action: Allowing normal traffic

Source: demo-client | Strikes: 0

EdgeShield is continuously inspecting hardware counter windows at the edge and is ready to mitigate suspicious cache behavior before it reaches the application.

Attack Rate (24h)

0% stable

Edge Latency

0 ms inference time

Avg Confidence

0% certainty

Real-Time Threat Timeline

Streaming inference from hardware counters

Benign Confidence
Attack Score
Attack Detected

Why was this flagged?

Explainability for latest decision

waiting

Live Activity Feed

Recent edge decisions

0 events
TimePredictionConfidenceLatencyAction Taken

Why Edge Detection?

Cloudflare Workers vs Traditional Infrastructure

EdgeShield
~1 ms
Detection & mitigation occurs at the edge, blocking attacks before they reach the server.
Traditional Cloud
80-150 ms
Detection only happens after the malicious request reaches the backend infrastructure.

System Pipeline

01

HPC Data Stream

Client perf counters

02

Cloudflare Worker

Edge API receives payload

03

Detection Engine

Feature eng. & ML ruleset

04

Mitigation

Allow or Rate Limit

Benchmark Accuracy

0% 2,000-row eval

Attack Recall

0% attacks detected

False Positives

0% benign rows flagged

Evaluate Custom Dataset

Upload CSV or JSON built from the six required perf counters.

Required Columns
cache_misses, cache_references, instructions, cycles, branches, branch_misses, label
0
0%
0%

Confusion Matrix

Computed when ground-truth labels are present.

0
0
0
0

Cache Behavior

0% dominant signal

Branch Behavior

0% control-flow

Execution Timing

0% IPC & cycles

How the model works

Worker rebuilds 36 features and applies transparent ML ruleset.

01

Raw counters

misses, references, instructions, cycles, branches.

02

Engineered features

IPC, CPI, miss rates, rolling statistics.

03

Edge decision

Weighted rules vote toward attack or benign.

Feature Importance Details

Live contribution breakdowns across categories.