[Paper Review] A New Fuzzy Approach for Dynamic Load Balancing Algorithm

Abbas Karimi, Faraneh Zarafshan+3 more|ArXiv.org|Oct 2, 2009

Distributed and Parallel Computing Systems14 references28 citations

TL;DR

This paper proposes a novel fuzzy logic-based dynamic load balancing algorithm that addresses uncertainty and inconsistency in distributed system state information. By using fuzzy inference to evaluate processor loads and make real-time decisions, the approach reduces response time by 30.84% compared to round-robin and 45.45% compared to random allocation, demonstrating superior performance in dynamic environments.

ABSTRACT

Load balancing is the process of improving the Performance of a parallel and distributed system through is distribution of load among the processors [1-2]. Most of the previous work in load balancing and distributed decision making in general, do not effectively take into account the uncertainty and inconsistency in state information but in fuzzy logic, we have advantage of using crisps inputs. In this paper, we present a new approach for implementing dynamic load balancing algorithm with fuzzy logic, which can face to uncertainty and inconsistency of previous algorithms, further more our algorithm shows better response time than round robin and randomize algorithm respectively 30.84 percent and 45.45 percent.

Motivation & Objective

• To address the limitations of traditional load balancing algorithms in handling uncertainty and inconsistency in system state information.
• To improve response time and performance in dynamic distributed and parallel computing environments.
• To develop a fuzzy logic-based decision mechanism that adapts to real-time load variations more effectively than deterministic or random methods.
• To evaluate the proposed algorithm against established benchmarks such as round-robin and random allocation.

Proposed method

• The algorithm uses fuzzy logic to process crisp input values representing processor load metrics, such as CPU utilization and queue length.
• Fuzzy inference rules are defined to map input load states to appropriate load distribution decisions based on linguistic variables like 'low', 'medium', and 'high' load.
• The system dynamically evaluates the load state of each processor at runtime and redistributes tasks based on fuzzy output scores.
• The fuzzy logic controller is designed to minimize response time by prioritizing underloaded or balanced processors for new task assignment.
• The approach avoids reliance on precise numerical thresholds, making it robust to noisy or inconsistent sensor or monitoring data.

Experimental results

Research questions

• RQ1 How can fuzzy logic effectively model uncertainty in distributed system load states to improve load balancing decisions?
• RQ2 To what extent does a fuzzy-based dynamic load balancing algorithm outperform traditional methods like round-robin and random allocation in terms of response time?
• RQ3 Can fuzzy inference reduce sensitivity to inconsistent or imprecise load monitoring data in real-time distributed systems?

Key findings

• The proposed fuzzy load balancing algorithm achieves a 30.84% reduction in response time compared to the round-robin algorithm.
• The algorithm reduces response time by 45.45% compared to the random allocation method, demonstrating significant performance gains.
• The fuzzy approach effectively handles uncertainty and inconsistency in system state information, improving decision robustness.
• The algorithm shows consistent performance improvements across dynamic workloads, indicating strong adaptability.