A cost-effective strategy for splitting and allocating alerts' workloads during forensic investigations of very large IDS logs

Abstract

The scarcity of suitable models that investigators can adopt to effectively manage and allocate resources during an outbreak of intrusions is gaining momentum in digital forensics. This paper proposes a cost-effective model that employs two measures of impurity to split and allocate alerts' workloads to two or more investigators of large intrusion logs. The premise is that the splitting and allocation of alerts' workloads to multiple investigators should be carried out such that the organisation will not incur marginal cost. The process must also guarantee timely delivery of forensic results whenever the intrusion detection systems (IDSs) trigger large quantity of alerts that exceed the capability of one investigator to analyse. This model uses large intrusion datasets, C++ language, clustering, entropy and Gini Index to demonstrate the profitable split-threshold that minimizes costs. The model pragmatically justifies the inclusion or exclusion of certain investigators of intrusion logs during critical moments at workplace.