Elsevier

Computer Networks

Volume 54, Issue 4, 19 March 2010, Pages 573-588
Computer Networks

A unified model for joint throughput-overhead analysis of random access mobile ad hoc networks

https://doi.org/10.1016/j.comnet.2009.08.019Get rights and content

Abstract

An analytical framework is developed to study the throughput and routing overhead for proactive and reactive routing strategies in random access mobile ad hoc networks. To characterize the coexistence of the routing control traffic and data traffic, the interaction is modeled as a multi-class queue at each node, where the aggregate control traffic and data traffic are two different classes of customers of the queue. With the proposed model, the scaling properties of the throughput, maximum mobility degree supported by the network and mobility-induced throughput deficiencies are investigated, under both classes of routing strategies. The proposed analytical model can be extended to evaluate various routing optimization techniques as well as to study routing/relaying strategies other than conventional proactive or reactive routing. The connection between the derived throughput result and some well-known network throughput capacity results in the literature is also established.

Introduction

Mobile ad hoc networks (MANETs) support a variety of new applications in many military and civilian settings due to the availability of portable wireless communication devices and the flexibility offered when networking them. A fundamental research question is the capability of supporting data traffic in large-scale MANETs, i.e., how feasible a large-scale MANET is. Many theoretical results have been recently discovered for the extreme performance points of a MANET under certain assumptions, e.g. [1], [2], [3], [4], [5]. All of these works focus on characterizing the fundamental scaling properties of the performance of a MANET without restriction of the choice on routing/relaying scheme used in the network – typically an idealized scheme is assumed. In parallel, various practical routing protocols have been proposed for MANETs (e.g. see [6], [7] and references therein). The performance of these routing protocols has been primarily evaluated by simulations. Simulation results reported in the literature show that the performance of a routing protocol heavily depends on several key parameters such as node density, mobility degree, traffic pattern, to name a few. However, simulation itself as a tool is limited in that it provides no analytical expressions of the impacts of one or more of these parameters on the performance. Simulations also do not scale well. This work fills in this gap.

Thus, the key question that this work addresses is: Given a particular routing strategy, how does the achievable performance scale with the network parameters?

In this paper, we present a framework for addressing this question and provide answers for particular classes of routing protocols of MANETs: proactive routing and reactive routing. The number of nodes and the mobility degree in the network are two of the most important network parameters in evaluating the feasibility of a routing strategy in a large-scale dynamic wireless network. For the mobility metric, we use the average relative speed in the network, which has been empirically shown to be a good metric to quantify the dynamics of a MANET [8].

Unlike some of the recent theoretical works in studying the achievable throughput in a MANET that assume that mobility can be used for improving throughput, such as [2], [3], [4], [5], mobility actually plays an opposite role in many networks, such as in situations with localized mobility, or mobility with variance that is much lower than the tolerable application delay. It is well-known that under these conditions, mobility does not improve throughput. For example, in proactive routing, mobility-induced link breakages initiate routing-layer actions to update and propagate the topology information. In reactive routing, mobility-induced path breakages initiate routing-layer actions for repairing or re-discovering path(s) between the corresponding source–destination pair(s). In either case, the control overhead is generated and propagated over the network. (In this paper, the terms of routing overhead, control overhead and control traffic are exchangeable.) Such control traffic consumes a portion of network resources and thus affects the achievable throughput of the network. This issue has not been sufficiently considered in the theoretical literature.

Thus, in this paper, the coexistence of such mobility-induced control traffic with the user-generated (i.e. applications) data traffic in the network motivates the development of a generic multi-class queue model at a node, where the aggregate control traffic and aggregate data traffic are two different classes of customers of the queue. From the stability requirement of a queue, we first analyze the throughput of the network under both proactive and reactive routing strategies and characterize its scaling properties. Then, we study the impact of mobility on the throughput performance of the network from two metrics: (i) the critical degree of mobility beyond which all the capacity of the network is consumed by control traffic; and (ii) the mobility-induced throughput deficiency which quantifies the negative effect of mobility on the throughput of proactive/reactive routing. The analytical results are validated by simulations. We further show some extensions of the proposed analytical model. On one hand, the model can be readily extended to evaluate the effectiveness of various routing optimization techniques in a large-scale MANET, such as expanding ring search [9], route caching [10] and multi-point relaying [11]. On the other hand, the proposed analytical model can be generalized to analyze the routing/relaying strategies other than conventional proactive or reactive routing, for example, the routing strategies with geographical location information [12]. We also discuss the connection between the derived throughput result and well-known network throughput capacity results in [1], [2].

The rest of the paper is organized as follows. Section 2 introduces the network model, including the topology and traffic setting, routing strategies and the proposed queue model. Section 3 analyzes the mobility-induced control traffic rates in both proactive and reactive routing. Section 4 proposes a random access MAC layer model and based on this MAC model, the service rates of the proposed queue model are derived. Section 5 presents a detailed analysis in throughput and routing overhead and the effectiveness of analytical results is verified by simulation in Section 6. The discussions and extensions on the proposed analytical model are presented in Section 7. Related work is summarized in Section 8. Section 9 concludes the paper.

Section snippets

Topology and traffic setting

We consider n+1 mobile nodes randomly distributed in a plane of unit area, with a torus border rule1 [13]. The position of any node is assumed to be a stationary random process with stationary distribution uniform on the area of interest and the trajectories of different nodes are independently and identically distributed; each node is equipped with an omnidirectional

The mobility-induced control traffic in proactive routing

In proactive routing, the amount of mobility-induced control traffic at any node is determined by the link change rate λc,l and the rate of transmitting “HELLO” messages λc,h. We determine these two quantities as follows.

To derive the link change rate λc,l at an arbitrary node A, we consider the (possible) link between node A and another arbitrarily chosen node B. Since both node A and B are mobile, we consider the movement of node B relative to node A. We assume that the relative movement of

A service model

The (transmission) service for a data/control packet at a node is under the control of the MAC layer protocol. Random access is one of the most popular MAC strategies in large-scale MANETs, due to its simplicity and robustness in dynamic environments. We adopt a random access MAC protocol model, similar to that in [17]. The basic operation of this MAC model is:

  • 1.

    Before transmitting each packet (either a control packet or a data packet), a node counts down a random back-off timer; the duration of

Throughput analysis

In this section, we use the proposed queue model to characterize the scaling properties of throughput and overhead in proactive routing and reactive routing, respectively.7 Specifically, the scaling result of the throughput per source–destination pair (i.e., per session), the

Simulation validation

We validate the analytical results obtained in Section 5 against simulations. For the scaling properties of the throughput per source–destination pair and the maximum mobility degrees supported by the network, we use MATLAB to perform Monte Carlo simulations in large scale networks. For the linear relationship between the reciprocal of relative throughput deficiency and the reciprocal of average relative speed given in (40), we carry out a packet level simulation in a network with 40 nodes,

Discussion and extensions of the model

In this section, we discuss several extensions of the proposed analytical model. The discussion shows how the proposed model is extended to study the scaling properties of the throughput and overhead with various routing strategies and optimization techniques in large-scale MANETs. The main results of this section are summarized in Table 2.

Related work

The limits of supporting data traffic in large-scale MANETs have been the focus of significant research interest in recent years. On one hand, efforts have been put into characterizing the fundamental scaling properties of the performance of a MANET without restriction on the choice of routing/relaying scheme used in the network. Following the well-known capacity results for static wireless networks by Gupta and Kumar [1], Grossglauser and Tse have analyzed the capacity of a wireless ad hoc

Conclusions

We have proposed an analytical framework to study the throughput and routing overhead for practical routing strategies in random access MANETs. By considering the coexistence of mobility-induced control traffic and data traffic at any node in the network, we have modeled the individual node as a generic multi-class queue. The analysis has shown that (i) the throughput per source–destination pair for both proactive routing and reactive routing is OWnlogn; (ii) there is a strong linear

Zhenzhen Ye received the B.E. degree from Southeast University, Nanjing, China, in 2000, the M.S. degree in high performance computation from National University of Singapore, Singapore, in 2003, and the M.S. degree in electrical engineering from University of California, Riverside, CA in 2005. He is currently working towards the Ph.D. degree in electrical engineering in Rensselaer Polytechnic Institute, Troy, NY. His research interests lie in the areas of wireless communications and

References (26)

  • M. Abolhasan et al.

    A review of routing protocols for mobile ad hoc networks

    Ad Hoc Networks

    (2004)
  • N. Bisnik et al.

    Queuing network models for delay analysis of multihop wireless ad hoc networks

    Ad Hoc Networks

    (2009)
  • P. Gupta et al.

    The capacity of wireless networks

    IEEE Trans. Inform. Theory

    (2000)
  • M. Grossglauser et al.

    Mobility increases the capacity of ad hoc wireless networks

    IEEE/ACM Trans. Networking

    (2002)
  • G. Sharma et al.

    Delay and capacity trade-offs in mobile ad hoc networks: a global perspective

    IEEE/ACM Trans. Networking

    (2007)
  • N. Bansal, Z. Liu, Capacity, delay and mobility in wireless ad-hoc networks, in: Proceedings of 22nd IEEE Annual Joint...
  • J. Mammen et al.

    Throughput and delay in random wireless networks with restricted mobility

    IEEE Trans. Inform. Theory

    (2007)
  • E. Royer et al.

    A review of current routing protocols for ad hoc mobile wireless networks

    IEEE Personal Commun.

    (1999)
  • F. Bai et al.

    Modeling path duration distributions in MANETs and their impact on reactive routing protocols

    IEEE J. Select. Area. Commun.

    (2004)
  • C. Perkins, E. Royer, S.R. Das, Ad hoc on-demand distance vector (AODV) routing draft-ietf-manet-aodv-13, Mobile Ad Hoc...
  • David B. Johnson et al.

    DSR: the dynamic source routing protocol for multi-hop wireless ad hoc networks

  • A. Laouiti, A. Qayyum, L. Viennot, Multipoint relaying: an efficient technique for flooding in mobile wireless...
  • M. Mauve et al.

    A survey on position-based routing in mobile ad hoc networks

    IEEE Network

    (2001)
  • Cited by (7)

    • Measurement and modelling of TCP downstream throughput dependence on SNR in an IEEE802.11b WLAN system

      2018, Journal of King Saud University - Engineering Sciences
      Citation Excerpt :

      The hidden node problem occurs when a station attempts to access and use network resources (transmit its own data) because it cannot sense that another station with a weak signal (usually far away from the WLAN radio or the station attempting to transmit its data) is already transmitting its data hence packet collision will occur which degrades the throughput observed (Hung and Bensaou, 2011). Network designers are constrained to continue to use pinging because available network throughput estimation tools such as Tamosoft throughput test, QCHECK, IxChariot, etc. require lots of time and network resources while available off network throughput models require lots of parameters to be specified (Padhye et al., 2000; Moltchanov, 2010; Gupta et al., 2011; Zhu et al., 2012; Wu et al., 2011; Detti et al., 2011; Loiseau et al., 2010; Ye and Abouzeid, 2010; Hung and Bensaou, 2011; Panda and Kumar, 2012; Tian and Tian, 2012). Since Network designers presently proceed with WLAN design by measuring received signal strength indication (RSSI) at different positions, can throughput be modelled as functions of SNR only within reasonable accuracy?

    • Performance reliability evaluation for mobile ad hoc networks

      2018, Reliability Engineering and System Safety
      Citation Excerpt :

      The routing protocol is assumed to be in a single-path and on-demand fashion, which is often used in practical applications [36]. A path found by the routing protocol follows the shortest path, and the length of a path is the number of links forming the path [27]. Carrier sense multiple access with collision avoidance (CSMA/CA) is employed as the medium access control (MAC) protocol [37].

    • An efficient metric for reliable routing with link dependencies

      2012, IEEE Vehicular Technology Conference
    • Practical link reliability for ad hoc routing protocol

      2011, IEEE Vehicular Technology Conference
    • Comparison of redundancy strategies in ad hoc networks

      2011, CFIP 2011 - Colloque Francophone sur l'Ingenierie des Protocoles
    View all citing articles on Scopus

    Zhenzhen Ye received the B.E. degree from Southeast University, Nanjing, China, in 2000, the M.S. degree in high performance computation from National University of Singapore, Singapore, in 2003, and the M.S. degree in electrical engineering from University of California, Riverside, CA in 2005. He is currently working towards the Ph.D. degree in electrical engineering in Rensselaer Polytechnic Institute, Troy, NY. His research interests lie in the areas of wireless communications and networking, including stochastic control and optimization for wireless networks, cooperative communications in mobile ad hoc networks and wireless sensor networks, and ultra-wideband communications.

    Alhussein A. Abouzeid received the B.S. degree with honors from Cairo University, Cairo, Egypt in 1993, and the M.S. and Ph.D. degrees from University of Washington, Seattle, WA in 1999 and 2001, respectively, all in electrical engineering. From 1993 to 1994, he was with the Information Technology Institute, Information and Decision Support Center, The Cabinet of Egypt, where he received a software engineering diploma. From 1994 until 1997, he served as a Project Manager in the Middle East Regional Office of Alcatel telecom, designing integrated voice/data enterprise solutions, and responsible for the technical launch of new Alcatel enterprise and public data switches. During his Ph.D. study, he held summer appointments with Allied Signal -now Honeywell- Redmond, WA, in 1999 and Hughes Research Labs, Malibu, CA, in 2000. In 2001 he joined Rensselaer Polytechnic Institute (RPI), Troy, NY, where he is currently an associate professor in the Electrical, Computer and Systems Engineering Department, and Deputy Director of the Center for Pervasive Computing and Networking, which he co-founded. Since December 2008, he has been Program Director (on leave from RPI), Directorate for Computer & Information Science & Engineering, Division of Computer & Network Systems, National Science Foundation, Arlington, VA, where he is responsible for the Networking Technology and Systems program. He received the NSF Faculty Early Career Development Award (CAREER) in 2006. His research focuses on protocols for dynamic networks. He serves on various conferences organization committees and is an Area Editor for Elsevier Computer Networks journal. He is an active member of IEEE and ACM.

    View full text