Nature-inspired survivability: Prey-inspired survivability countermeasures for cloud computing security challenges

As cloud computing environments become complex, adversaries have become highly sophisticated and unpredictable. Moreover, they can easily increase attack power and persist longer before detection. Uncertain malicious actions, latent risks, Unobserved or Unobservable risks (UUURs) characterise this new threat domain. This thesis proposes prey-inspired survivability to address unpredictable security challenges borne out of UUURs. While survivability is a well-addressed phenomenon in non-extinct prey animals, applying prey survivability to cloud computing directly is challenging due to contradicting end goals. How to manage evolving survivability goals and requirements under contradicting environmental conditions adds to the challenges. To address these challenges, this thesis proposes a holistic taxonomy which integrate multiple and disparate perspectives of cloud security challenges. In addition, it proposes the TRIZ (Teorija Rezbenija Izobretatelskib Zadach) to derive prey-inspired solutions through resolving contradiction. First, it develops a 3-step process to facilitate interdomain transfer of
concepts from nature to cloud. Moreover, TRIZ’s generic approach suggests specific
solutions for cloud computing survivability. Then, the thesis presents the conceptual prey-inspired cloud computing survivability framework (Pi-CCSF), built upon TRIZ derived solutions. The framework run-time is pushed to the user-space to support evolving survivability design goals. Furthermore, a target-based decision-making technique (TBDM) is proposed to manage survivability decisions. To evaluate the prey-inspired survivability concept, Pi-CCSF simulator is developed and implemented. Evaluation results shows that escalating survivability actions improve the vitality of vulnerable and compromised virtual machines (VMs) by 5% and dramatically improve their overall survivability. Hypothesis testing conclusively supports the hypothesis that the escalation mechanisms can be applied to enhance the survivability of cloud computing systems. Numeric analysis of TBDM shows that by considering survivability preferences and attitudes (these directly impacts survivability actions), the TBDM method brings unpredictable survivability information closer to decision processes. This enables efficient execution of variable escalating survivability actions, which enables the Pi-CCSF’s decision
system (DS) to focus upon decisions that achieve survivability outcomes under unpredictability imposed by UUUR.