DISTRIBUTED COMPUTING LEADER ELECTION

Distributed Computing Leader Election in Different Topologies

Distributed Computing Leader Election in Different Topologies

Introduction

In a distributed computing system, a change in the state of the system usually requires that the new state be propagated to all active nodes so that they can rightly perform cooperative tasks. Since more than one node may notice a status change, conflicts in state determination and propagation are possible. In order to double-check that all nodes of a communicating group are at the same state, all nodes of the group must acquiesce upon the election of a lone leader (coordinator). In this study paper, we present an Election Protocol for distinct configuration systems. The protocol was tested on distinct well renowned configurations and its performance was studied utilising replication modeling techniques. The simulation study displayed a tradeoff between the election time and the number of swapped messages. A technique to minimize the number of messages was then introduced. Both positive and negative effects of this technique were observed (Lynch,1996).

Discussion and Analysis

Leader election is a very significant problem, not only in wired networks, but in mobile publicity hoc systems as well. However, due to the node mobility and network performance, cluster foremost node may be lost which will affect the communication severely. According to the characteristics of poor mobility of low-level node, we present a low-level asynchronous algorithm with the collision avoidance mechanism, and explained the problem above. And we designed experiments in application layer. The leader-elected time was got from those experiments and the outcome was in evaluation with synchronous algorithm. This algorithm can be utilised in blaze battling, construction supervision and rescue work in which a hierarchy ad hoc network is needed to dynamically established. By electing a new foremost in case of preceding leader loss, the algorithm can enhance the robustness of whole network.

The most fundamental observation with an significant outcome is that backoff delays not only avoid collision, but furthermore provide a precious opening to prioritize distinct nodes, endowing a straightforward solution to the local leader election problem. A widespread implicit synchronization issue pursued by distinct backoff delays assigns distinct possibilities of evolving the leader to all participating nodes. Each node, after discerning the implicit synchronization point, calculates a backoff delay founded on a certain criterion and then groups a backoff timer with duration identical to the backoff delay. When the backoff timer expires, the node can start transmitting an announcement packet. However, if the node receives an announcement packet from another node before its own backoff time expires, it cancels the backoff timer. Thus, in most situations, only the node with the least significant backoff delay will do well in transmitting the announcement packet, and routinely will become the local leader. Upon receiving the announcement packet, other nodes know that a new local leader has been elected (Barbosa, 1996).

Now, we presents a different scalable leader election protocols for large method assemblies with a weak membership requirement. The protocols trade correctness for scale, that is, it presents very good ...