Table 1 Complexity analysis of proposed classification and mitigation algorithm.
Step | Operation | Time complexity | Space complexity | Remarks |
|---|---|---|---|---|
1 | Threat probability estimation (per VM) | O(1) | O(1) | Direct lookup or ML model inference cost. |
2 | Policy influence matrix update | O(A) | O(A) | Iterates over actions (A). |
3 | Normalization of policy probabilities | O(A) | O(A) | Requires summing and scaling probabilities. |
4 | Action sampling (categorical) | O(A) | O(1) | Uses probability distribution; negligible storage. |
5 | Expected cost computation | O(A) | O(A) | Each action’s cost is evaluated. |
6 | Optimal action selection | O(A) | O(1) | Linear scan over expected costs. |
7 | Threshold-based execution decision | O(1) | O(1) | Simple comparison. |
8 | Recovery index update | O(1) | O(1) | State-dependent constant operation. |
9 | Adaptive learning rate adjustment | O(1) | O(1) | Threshold-based update. |
10 | Adaptive threshold adjustment | O(1) | O(1) | Minor threshold modification. |
11 | Entropy calculation of action history | O(A) | O(A) | Requires iterating over action probabilities. |
12 | Exploration of new actions | O(A) | O(A) | Search among unexplored actions. |
13 | Cooldown enforcement | O(1) | O(1) | Timestamp comparison. |
14 | Global anomaly feedback update | O(N) | O(N) | Aggregates scores across VMs (N). |
