Exploiting Dynamic Querying like Flooding Techniques in Unstructured Peer-to-Peer Networks
Hongbo Jiang, Shudong Jin
Abstract
Hongbo Jiang, Shudong Jin
Abstract
In unstructured peer-to-peer networks, controlled flooding aims at locating an item at the minimum message cost. Dynamic querying is a new controlled flooding technique. While it is implemented in some peer-to-peer networks, little is known about its undesirable behavior and little is known about its general usefulness in unstructured peer-to-peer networks. This paper describes the first evaluation and analysis of such techniques, and proposes novel techniques to improve them. We make three contributions. First, we find the current dynamic querying design is flawed. Although it is advantageous over the expanding ring algorithm in terms of search cost, it is much less attractive in terms of peer perceived latency, and its strict constraints on network connectivity prevent it from being widely adopted. Second, we propose an enhanced flooding technique which requires the search cost close to the minimum, reduces the search latency by more than four times, and loosens the constraints on the network connectivity. Thus, we make such techniques useful for the general unstructured peer-to-peer networks. Third, we show that our proposal requires only minor modifications to the existing search mechanisms and can be incrementally deployed in peer-to-peer networks.
OpenAlex reports 56 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In unstructured peer-to-peer networks, controlled flooding aims at locating an item at the minimum message cost. Dynamic querying is a new controlled flooding technique. While it is implemented in some peer-to-peer networks, little is known about its undesirable behavior and little is known about its general usefulness in unstructured peer-to-peer networks. This paper describes the first evaluation and analysis of such techniques, and proposes novel techniques to improve them. We make three contributions. First, we find the current dynamic querying design is flawed. Although it is advantageous over the expanding ring algorithm in terms of search cost, it is much less attractive in terms of peer perceived latency, and its strict constraints on network connectivity prevent it from being widely adopted. Second, we propose an enhanced flooding technique which requires the search cost close to the minimum, reduces the search latency by more than four times, and loosens the constraints on the network connectivity. Thus, we make such techniques useful for the general unstructured peer-to-peer networks. Third, we show that our proposal requires only minor modifications to the existing search mechanisms and can be incrementally deployed in peer-to-peer networks.
Key concepts: Flooding (psychology), Computer science, Peer-to-peer, Latency (audio), Distributed computing, Computer network, Telecommunications, Psychology