Скачати 8.01 Kb.
DYNAMIC LOAD BALANCING IN CONJUNCTION WITH
ADAPTIVE AQM ALGORITHM
Olena Slavko, Inna Shapoval
Kremenchuk Mykhailo Ostrohradskyi National University
Intensive development of advanced network technologies (technologies of wireless and Ad Hoc networks) is conditioned by specific of services provided for Internet users. Online watching videos and listening to audio content, interactive games, IP-telephony, Internet conferences require providing of appropriate level of Quality of Service (QoS). This task is complicated by significant amount and heterogeneity of traffic transmitted in real time in case of uncertainty of data flow dynamic and network environment.
Optimization methods of service software, data caching, scaling of system's hardware and methods of congestion avoidance and congestion control are used to solve the problem of congestions in network nodes or their groups. Methods for control of network resources also include dynamic workload balancing of nodes in computer networks. Broadcast requests that are used by politics of determination the possibility of nodes' participation in load balancing, result in a significant increase of network traffic and congestions. This increase the load imbalance of network nodes.
In the paper is offered to join the modified algorithm of adaptive random early detection (ARED), developed in , and the adaptive control algorithm of network traffic based on fuzzy logic and active networks concepts, developed in . ARED belongs to methods of active queue management (AQM), realized in router buffers.
Modified ARED uses a corrective parameter in definition of average queue length in router buffer based on analysis of stability conditions of queue and dynamic setting of maximum probability of packets dropping in case of congestions. Adaptive load balancing algorithm controls data flow based on changes in the intensity balancing based on a current state of network links load.
Combination of these methods improves a network bandwidth and prevent congestions at high intensity of network traffic by using fuzzy logic to increase the quality of load balancing, as well as by reducing the time setting of control parameters of control algorithms for network data flows in real time to QoS providing for heterogeneous traffic.
|Theme of №1 Compensatory and adaptive processes: hypertrophy, hyperplasia||Teaching load exam|
|Teaching load exam||Experimental dynamic system identification of damaged beam volkova V|
|Endurance of centrally pressed concrete elements joined by acrylic glues at dynamic loads of different frequencies||Diagnostic Algorithm for Sexual Precocity|
|Diagnostic Algorithm for Sexual Precocity||Diagnostic Algorithm for Diseases with Polyuria and Polydipsia|
|Diagnostic Algorithm for Haemorrhagic Syndrome in Children||Diagnostic Algorithm for Diseases with Polyuria and Polydipsia|