Multi-user multi-objective computation offloading for medical image diagnosis.

Area of Science:

• Computer Science
• Medical Informatics
• Distributed Systems

Background:

• Computation offloading addresses limited computing resources in medical devices by utilizing edge servers for tasks like image diagnosis.
• Existing offloading strategies often overlook crucial factors such as individual user risk awareness and diverse processing costs.
• Efficient computation offloading is vital for timely and accurate medical image analysis.

Purpose of the Study:

• To propose a multi-user, multi-objective computation offloading strategy for medical image diagnosis.
• To incorporate user risk awareness, delay, energy consumption, and payment into a utility function.
• To develop a low-complexity algorithm that optimizes offloading decisions for enhanced diagnostic efficiency.

Main Methods:

• Designed a prospect theoretic utility function integrating delay, energy, payment, and risk awareness.
• Formulated the offloading problem as a distributed optimization problem aimed at maximizing user utility.
• Transformed the problem into a non-cooperative game, proving Nash equilibrium points using exact potential game theory.
• Developed a low-complexity computation offloading algorithm based on best response dynamics.

Main Results:

• The proposed algorithm demonstrates faster convergence compared to benchmark and heuristic methods.
• The algorithm achieves a minimal 1.14% decrease in utility value even with an increasing number of users.
• Numerical experiments validated the algorithm's performance and the impact of various parameters on utility.

Conclusions:

• The developed computation offloading strategy effectively balances multiple objectives in medical image diagnosis.
• The algorithm provides an efficient and scalable solution for resource-constrained medical environments.
• This approach enhances the utility of computation offloading for medical image diagnosis by considering user-specific factors.