Tracking and Predicting End-to-End Quality in Wireless Community Networks

Abstract

Abstract-Community networks are an emergent model with mottos like "a free net for everyone is possible" or "don't buy the network, be the network". Their social impact is measurable, as the community is provided with the right and opportunity of communication. The combination of wired and wireless links in these networks, and the unreliable nature of the wireless medium, poses several challenges to the routing protocol. End-toEnd quality tracking helps the routing layer to select paths that maximize the delivery rate and minimize traffic congestion. We believe that End-to-End quality prediction can be a technique that surpasses End-to-End quality tracking by foreseeing which paths are more likely to change quality. In this work, we focus on End-to-End quality prediction by means of time-series analysis. We apply this prediction technique in the routing layer of largescale, distributed, and decentralized networks. We demonstrate that it is possible to accurately predict End-to-End Quality with an average Mean Absolute Error of just 2.4%. Particularly, we analyze the path properties and path ETX behavior to identify the best prediction algorithm. Moreover, we analyze the EtEQ prediction accuracy some steps ahead in the future and also its dependency of the time of the day.

Time Series Analysis to Predict End-to-End Quality of Wireless Community Networks

“…Moreover, the EtEQ prediction accuracy is analyzed some steps ahead in the future and also its dependency of the time of the day. In this work, we extend our previous work [19] but including several new analysis which would allow a better understanding of End-to-End quality prediction in Wireless Mesh Community Networks. Particularly, we provide: (1) a detailed analysis of path properties behavior showing that Path Quality prediction is possible and meaningful; (2) an analysis of the computational cost of the prediction in terms of CPU utilization, energy consumption and temperature to confirm that the prediction can be executed directly in the routers; and (3) a detailed analysis of an use case that shows the behavior of two orthogonal routing strategies, in terms of Packet Delivery Ratio, ETX of links, and End-to-End Delay.…”

“…Finally, Millán et al [19] focused on End-to-End quality prediction by means of time-series analysis. This prediction technique is applied in the routing layer of large scale, distributed, and decentralized networks.…”

A Survey of Anticipatory Mobile Networking: Context-Based Classification, Prediction Methodologies, and Optimization Techniques

“…As an example, the inter-download times of video segments are predicted in [102], where the output sequences are the interdownload times of the already downloaded segments and the states are the instants of the next download request. ARIMA: [13], [38], [40], [46], [47], [54], [58], [59], [63], [100], [119] Kalman: [32], [ CF: [16], [134], [149] Cluster: [15], [34], [51], [117], [122], [123], [148], [156] Decision trees: [35], [98], [ Functional: [28], [29], [38], [64], [99], [104], [105] SVM: [51], [114], [139] ANN: [14], [48], [106], [ 2) Bayesian inference: This approach allows to make statements about what is unknown, by conditioning on what is known. Bayesian prediction can be summarized in the following steps: 1) define a model that expresses qualitative aspects of our knowledge but has unknown parameters, 2) specify a prior probability distribution for the unknown parameters, 3) compute the posterior probability distribution for the parameters, given the observed data, and 4) make predictions by averaging ove...…”

Predicting Topology Propagation Messages in Mobile Ad Hoc Networks: The Value of History

“…Recently, time-series analysis has been considered to estimate link quality (LQ) and end-to-end quality (EtEQ) in the routing layer, involving real-world wireless mesh community networks. For instance, Millan et al [19][20][21][22] shown that time-series analysis can be used to improve the performance of the routing protocol, by providing information that allows making appropriate and timely decisions. This contributes to maximize the message delivery rate and minimize traffic congestion at both levels (i.e., LQ and EtEQ) with a small average mean absolute error.…”

“…However, the use of predictors increases the complexity of the routing protocols, because of the additional hardware and software required to make and validate predictions. Moreover, penalty mechanisms are usually introduced to the system when there is a high rate of mispredictions, which negatively affect the performance of these protocols.Prediction mechanisms have been embedded in routing protocols to foresee several aspects of a network, such as nodes mobility [13], reliability of its topology [14,15] and quality of links and end-to-end paths [16][17][18][19][20][21][22][23][24]. These mechanisms have also been used to reach particular communication…”