Accelerating Approximate Consensus with Self-Organizing Overlays
نویسنده
چکیده
Laplacian-based approximate consensus algorithms are an important and frequently used building block in many distributed applications, including formation control, sensor fusion, and synchronization. These algorithms, however, converge extremely slowly on networks that are more than a few hops in diameter and where values are spatially correlated. This paper presents a new algorithm, Power Law Driven Consensus, which uses a self-organizing virtual overlay network to accelerate convergence, at a cost of decreasing the predictability of the final converged value. Experimental comparison with the Laplacian approach confirms that PLD-consensus allows for drastically faster convergence in spatial networks.
منابع مشابه
Self-Organizing Overlays
Overlays are an important tool for deploying new services on the Internet. However, the existing overlays are statically configured. For fault tolerance, ease of management, and efficiency it is crucial that overlays be self-organizing. We present a distributed algorithm to automatically configure the nodes into a logical topology which is dynamic and efficient.
متن کاملLaplacian-based consensus on spatial computers
Robotic swarms, like all spatial computers, are a challenging environment for the execution of distributed consensus algorithms due to their scale, diameter, and frequent failures. Exact consensus is generally impractical on spatial computers, so we consider approximate consensus algorithms. In this paper, we show that the family of self-organizing protocols based on the graph Laplacian of a ne...
متن کاملScalable Self-Organizing Overlays
Recent developments including peer-to-peer networks and application-level multicast have generated interest in overlays, and by extension in algorithms for automatically maintaining overlay structures. In this paper, we present a self-organizing protocol for forming and managing overlays that scales to at least tens of thousands of nodes. Previous algorithms target small scale overlays and do n...
متن کاملTrading accuracy for speed in approximate consensus
Approximate consensus is an important building block for distributed systems, used overtly or implicitly in applications as diverse as formation control, sensor fusion, and synchronization. Laplacian-based consensus, the current dominant approach, is extremely accurate and resilient, but converges slowly. Comparing Laplacian-based consensus to exact consensus algorithms, relaxing the requiremen...
متن کاملSpatial Computing 2013 colocated with AAMAS Saint - Paul , 6 May 2013
Laplacian-based approximate consensus algorithms are an important and frequently used building block in many distributed applications, including formation control, sensor fusion, and synchronization. These algorithms, however, converge extremely slowly on networks that are more than a few hops in diameter and where values are spatially correlated. This paper presents a new algorithm, Power Law ...
متن کامل