Elsevier

Computers & Electrical Engineering

Volume 56, November 2016, Pages 130-144
Computers & Electrical Engineering

Dynamic traffic engineering for high-throughput data forwarding in wireless mesh networks

https://doi.org/10.1016/j.compeleceng.2016.08.004Get rights and content

Abstract

High-throughput data delivery in Wireless Mesh Networks (WMNs) is a challenging problem due to dynamic changes of link quality, interference and congestion. In this work, we first develop an optimization framework for Dynamic Traffic Engineering (O-DTE) in WMNs that aims to minimize the interference and congestion at each hop through joint power and rate control so as to achieve high-throughput data delivery. Due to NP-hardness of the O-DTE framework, we then develop a greedy heuristic alternate solution (G-DTE) that enables routers, at each hop, to select outgoing links offering higher data rates and reduced interferences. Thus, the proposed G-DTE produces near optimal results by taking multi-path data forwarding decisions in distributed fashion; it exploits single-hop neighborhood information only and thus it is scalable. The simulation results, carried out in ns-3, demonstrate that the proposed G-DTE significantly outperforms the state-of-the-art works in terms of throughput, delay, reliability and fairness performances.

Introduction

Wireless Mesh Network (WMN) has been recently emerged as a promising technology for wireless Internet infrastructure development because of its low cost, ease of deployment and installation [1]. The increasing number of users and diversified application usages as well as the incorporation of sensors and Internet of Things (IoT) devices with the WMNs has caused exponential growth in traffic flows [2]. This increased traffic volume, however causes congestion in the network, degrading application throughput and reliability and delay performances. Therefore, how to provide satisfactory network performance, by utilizing the limited offered bandwidth, has emerged as a problem. This paper explores traffic engineering policies that dynamically adjust transmission powers and data rates over outgoing links at each router so as to enhance the network throughput.

A good number of works in the relevant literature focus on the use of emerging wireless technologies including directional antennas [1], multiple-input-multiple-output (MIMO) [3] and multichannel and multiradio [4] solutions at lower layers for increasing data delivery throughput of a network. However, none of these approaches take into account the sudden surge of huge data traffic generated from diverse user (human or device) applications, that cause network to become congested. In WMNs, mesh routers act as intermediate nodes and forward user traffic to mesh gateways (GWs) in multi-hop fashion. A WMN with a single GW might create a bottleneck condition and a single point of failure for the network [5]. The use of multiple GWs in a WMN and multipath data forwarding towards them have been proven to provide better throughput performance [6] since this strategy increases the aggregated bandwidth for certain traffic flow.

There is a significant body of works that focus on multipath data forwarding to enhance flow performance in WMNs. Those studies include discovery of node-disjoint or link-disjoint multiple paths [7], optimal power allocation to mitigate interference [8], [9] and computing mutually interference-free multiple paths [6], [10]. However, these end-to-end data delivery systems are less responsive to dynamic network conditions (link quality, interference, traffic demand, etc.) and thus they utilize instantaneous network capacity poorly. A central controller is deployed in [11] to find collision-free scheduling for all forwarding links using a backtracking algorithm. However, it does not consider rate adjustment in presence of varying link conditions and traffic demands. Authors in [12] propose a cross layer design approach to joint traffic splitting, rate control, routing and scheduling that splits traffic over multiple paths to enhance flow performance. The employment of a centralized controller would increase scheduling delays, cause a single point of failure and lead to wrong scheduling decisions due to obsolete information usage.

To counteract the aforementioned deficiencies, a distributed control agent is required at each forwarding router to determine outgoing link qualities and split traffic over them in a way that can maximize network performance. In this paper, we propose an optimization framework for Dynamic Traffic Engineering (O-DTE) in WMNs that aims to minimize interference and congestion and thus enhances throughput performance. The framework adopted belongs to a mixed integer nonlinear programming (MINLP) problem and involves both combinatorial and continuous constraints, making it an NP-hard problem. A greedy heuristic alternate solution G-DTE has also been developed that produces near-optimal results. The key contributions of this paper are summarized as follows:

  • •

    The traffic forwarding policies of the proposed O-DTE and G-DTE systems minimize neighborhood interference and backlogged traffic, and explore the least congested next-hop nodes so that the overall throughput of the network is maximized.

  • •

    We define a new weight function for each link based on its current achievable rate and the congestion level of the downstream. We allow a G-DTE router to prioritize the forwarding links based on their weights and to select less congestive and high-throughput links for traffic splitting.

  • •

    Each G-DTE downstream router controls the weighted-fair reception of data traffic from its upstream nodes and a high level of fairness is maintained.

  • •

    The proposed G-DTE routers exploit single-hop neighborhood information only to take traffic forwarding decisions distributedly and so make it scalable.

  • •

    Finally, the results of our simulation experiments, carried out in ns-3 [13], show that the proposed G-DTE system offers significant performance improvements in terms of throughput, delay, reliability and flow fairness.

The rest of the paper is organized as follows. In Section 2, we explore the existing works in the relevant literature; while the system model is presented in Section 3. The objective function and constraints of the O-DTE framework are formulated in Section 4 and the G-DTE system is designed in Section 5. The performance evaluation results are presented in Section 6; while Section 7 concludes the paper.

Section snippets

Related works

A good number of works in the literature is focused on throughput improvement of WMNs by exploiting diverse aspects of the network including congestion mitigation [14], [15], rate adaptation [10], [16], scheduling [11], [12], channel allocation [17] and routing [16]; either independently or jointly [16]. Based on the number of paths over which a source node delivers data to a targeted destination, traffic forwarding paradigm appears to be either single path or multipath. A significant number of

System model and assumptions

Let a graph Γ=(V,E) represent a typical wireless mesh network, where V is the set of mesh nodes and E is the set of links; and, G⊂V is a set of mesh gateways. The clients are connected with edge routers and the core routers form a multihop network backbone that carries traffic mainly in between mesh clients and the Internet through gateways, as shown in Fig. 1. Each node v∈V has multiple radio interfaces, each operating on an orthogonal channel so as to facilitate enhanced network capacity by

Optimization framework for DTE

Our proposed optimal dynamic traffic engineering (O-DTE) at forwarding nodes aims to fulfill upstream flow demand while minimizing neighborhood interference and congestion. Let, for each link (vw)∈Lvd, the tuple (x(vw), p(vw)) represents the activation status and power allocation over link (vw) at any given time, where x(vw) ∈ {0, 1} and p∈P. Let Q(vw)={(x(vw),p(vw))}, be the set of all possible allocation vectors over each link (vw)∈Lvd and Sv=∏(vw)∈LvdQ(vw) be the set of all possible

Greedy heuristic solution to DTE

This section presents a greedy heuristic alternate solution G-DTE to the problem of DTE for high-throughput data forwarding problem that finds near optimal set of next-hop links and their power allocations so as to carry the aggregated upstream traffic flow by minimizing backlogged traffic.

Performance evaluation

In this section, we implement O-DTE, G-DTE, MRA [6], MRT [11] and CLC_DGS [12] in a discrete-event network simulator, ns-3 [13] and present the comparative performance results. In implementation of O-DTE, we have explored all possible assignments of power and data rates on all downstream links to find out the optimal allocation using the objective function of Eq. (4). However, for G-DTE implementation, we look for the first feasible set of links that satisfies the demand constraint from the

Conclusion

In this paper, we advance the study of high-throughput data forwarding strategies in Wireless Mesh Networks (WMNs) and explore two dynamic traffic engineering (DTE) approaches: optimal DTE (O-DTE) and greedy DTE (G-DTE). Since the O-DTE explores all possible power and rate allocations to available outgoing links, it maximizes the throughput via minimizing interference and congestion; however, it becomes an NP-hard problem. The G-DTE routers contribute to obtain near optimal throughput in

Acknowledgments

The authors would like to extend their sincere appreciation to the Deanship of Scientific Research at King Saud University for its funding of this research through the Research Group Project no. RGP-281.

Maheen Islam is a PhD student of the Department of Computer Science and Engineering, University of Dhaka, Bangladesh. She received her B.S. and M.S. degrees from the University of Dhaka in 1998 and 1999, respectively. Her areas of research interest include Wireless Mesh Networks, Energy Optimization, modeling and analysis of communication algorithms, etc. She is a student member of IEEE.

References (25)

  • T.C. A. Raniwala

    Architecture and algorithms for anieee -based multi-channel wireless mesh network

    IEEE INFOCOM

    (2005)
  • J. Jun et al.

    The nominal capacity of wireless mesh networks

    IEEE Wireless Commun

    (2003)

    • Cited by (11)

    • Mixed-integer bilinear and piecewise-linear models for designing switching strategies of multilevel power converters

      2019, Computers and Electrical Engineering
      Citation Excerpt :

      Alternatively, this paper proposes two formulations of the baseline problem that do not rely on trigonometric functions. The first is a Mixed-Integer Nonlinear Programming (MINLP) [10] formulation that does represent angles as variables but instead models their sine and cosine functions as independent variables. The resulting formulation is a consistent MINLP in which the nonlinear terms are all bilinear, thereby enabling the application of local and global MINLP algorithms.

    • Traffic engineering in cognitive mesh networks: Joint link-channel selection and power allocation

      2018, Computer Communications
      Citation Excerpt :

      Methods in [4,13], rely on centralized server to compute interference-free link scheduling and channel assignment to enhance flow throughput considering single-path data delivery between each source-destination pair. Due to instantaneously varying network conditions, decision execution at a central point leads to unstable system condition and the single path data delivery approach experiences restricted forwarding capacity [14]. On the other hand, multi-path data forwarding techniques can improve multimedia flow throughput, provide fault-tolerance routing and increase data security by splitting information over multiple paths between source and destination [15,16].

    • An adaptive traffic engineering approach based on retransmission timeout adjustment for software-defined networks

      2024, Journal of Ambient Intelligence and Humanized Computing

    • Resource Allocation Approach for Optimal Routing in IoT Wireless Mesh Networks

      2021, IEEE Access

    • Left-Right-Front Caching Strategy for Vehicular Networks in ICN-Based Internet of Things

      2021, IEEE Access

    • Markov transition and smart cache congestion control for IoT enabled wireless mesh networks

      2021, Peer-to-Peer Networking and Applications
    View all citing articles on Scopus

    Maheen Islam is a PhD student of the Department of Computer Science and Engineering, University of Dhaka, Bangladesh. She received her B.S. and M.S. degrees from the University of Dhaka in 1998 and 1999, respectively. Her areas of research interest include Wireless Mesh Networks, Energy Optimization, modeling and analysis of communication algorithms, etc. She is a student member of IEEE.

    Md. Abdur Razzaque obtained his PhD degree in Computer Engineering from Kyung Hee University, South Korea in 2009. He is a Professor of Computer Science and Engineering, University of Dhaka, Bangladesh. He has published more than 80 research papers in international conferences and journals and serves for many conferences and journals. He is a senior member of IEEE.

    Md. Mamun-Or-Rashid obtained his PhD degree in Computer Engineering from Kyung Hee University, South Korea. He is a Professor of Computer Science and Engineering, University of Dhaka, Bangladesh. His research interest is in the area of modeling, analysis and optimization of wireless networking protocols and architectures. He has published a good number of research papers in international conferences and journals.

    Mohammad Mehedi Hassan is an Assistant Professor of Information Systems Department, King Saud University, Riyadh, KSA. He received his Ph.D. degree in Computer Engineering from Kyung Hee University, South Korea in 2011. His research interests include Cloud collaboration, multimedia Cloud, sensor-Cloud, mobile Cloud, Thin-Client, Grid computing, IPTV, virtual network, sensor network, and publish/subscribe system.

    Ahmad Almogren is an Associate Professor of Information Systems Department, King Saud University, Riyadh, KSA. He received his MS in 1995 and Ph.D. in 2002 from Computer Science, Southern Methodist University, Dallas, Texas, USA. His research interests include mobile, cloud and pervasive computing, wireless sensor network, computer security, and data mining.

    Abdulhameed Alelaiwi is an Assistant Professor of Software Engg. Department, King Saud University. Riyadh, Saudi Arabia. He is currently the Vice Dean for Deanshhip of Scientific Research at King Saud University. He has authored many publications in the areas of software testing analysis and design, cloud computing, and multimedia.

    View full text
    © 2016 Elsevier Ltd. All rights reserved.