Elsevier

Computer Networks

Volume 50, Issue 3, 22 February 2006, Pages 367-397
Computer Networks

Performance evaluation of the push-mode-multicast based candidate access router discovery (PMM CARD)

https://doi.org/10.1016/j.comnet.2005.06.002Get rights and content

Abstract

In order to provide nomadic users with QoS-enabled services, advanced mobility management will prove to be of fundamental importance in the future Internet. It has been widely recognized that the basic Mobile IPv4/v6 protocols could perform poorly, especially with QoS-demanding applications. One of the main steps to achieve seamless handover is the quick discovery of the surrounding wireless coverage at each access router (AR), i.e., the discovery of IP addresses and service capabilities (SCs) of candidate access routers (CARs) to hand over to. The rapid knowledge of IP addresses allows mobile nodes (MNs) to speed-up the handover process, whereas information about SCs are important for the selection of the most appropriate wireless access (target access router, TAR) among the set of CARs, according to a given criterion (e.g., load balancing).

In this paper, we first describe the candidate access router discovery (CARD) solutions proposed within the framework of the IETF SEAMOBY WG, which has inspired research in this field. Then, we propose the distributed push-mode-multicast based CARD (PMM CARD) approach and compare it with the IETF proposals. The novelty of our solution is the use of push-mode multicast transmissions, which enables an efficient distribution of CARD information within the network, together with a significant reduction in latency due to explicit queries to a remote entity.

Then, we develop a theoretical model to compute the signaling burden associated with different CARD solutions. Our analysis shows that, even though the amount of signaling required to implement all CARD approaches is generally low, our approach gives a certain improvement over the IETF proposals in terms of the average signaling load. In addition, the results obtained by a simulation campaign show the effectiveness of the PMM CARD solution in terms of the time needed by ARs to discover the surrounding wireless coverage.

Introduction

Some of the latest novelties in the IP world have been introduced with the aim of harmonizing the inter-working of heterogeneous wired and wireless networks. The final goal of this trend is summarized by the frequently heard statement “information anywhere anytime”. Thus, what is needed is both a multipurpose network infrastructure and open service platforms, which allow such integration. From the networking perspective, the management of portable computing devices is referred to as Mobile Computing, which means to provide mobile terminals with a seamless, ubiquitous computing environment, through the merging of the recent advances in computing and communication technologies. In this framework, handover management is extremely important and challenging, particularly if it has to be seamless. In [1], the Authors define the seamless handover as “a handover in which there is no change in service capability, security, and quality”.

IP mobility protocols (i.e., Mobile IPv4/v6 [5], [6]) enable mobile nodes (MNs) to execute IP layer handover between access routers (ARs). It is well known that the basic Mobile IP (MIP) protocols perform poorly in supporting real time applications. A number of approaches have been proposed so far to improve MIP performance: (i) micromobility solutions [7], which aim to limit the scope of the MIP procedures, thus reducing handover latency and signaling burden; (ii) context transfer solutions [3], [4], the goal of which is to quickly re-establish information states (context) associated with the MN in the new AR upon handover; (iii) solutions that minimize packet loss and delay (fast handover and smooth handover) [10], [11].

All the proposed enhancements assume the awareness of the new AR to hand over to. The process to obtain this information is indeed the missing step, which plays a crucial role in the overall, seamless handover procedure. Thus, in advanced mobility scenarios, it is very important that an MN is able not only to discover the set of candidate ARs (CARs) to hand over to, but also to select the most appropriate one before handing over. For this reason, some candidate access router discovery (CARD) procedures have been recently proposed, with the aim of discovering and collecting the set of CARs, together with their service capabilities [2], [8], [9]. Through this information, it is possible to select the target AR (TAR) to hand over to, that optimizes a given metric.

The CARD requires two essential functions to be performed:

  • reverse address translation, which means to use layer 2 addresses to determine the coverage areas which can be served through a port of an AR, thus with its IP address (layer 3 coverage). This is important to speed-up MIP handovers, and is implemented by binding the layer 2 identifier (L2 ID) of each access point (AP) with the IP address of the CAR connected to it. MNs are assumed to listen to the L2 beacons transmitted by surrounding APs periodically and to learn their L2 IDs. This information is then exchanged with the current AR. The CARD procedures described below can then support MNs with layer 3 information. Thus, in order to execute the handover, MNs do not have to wait for the MIP router advertisement from the new AR and fast handover procedures may be executed [10], [11]. Note that the capacity of performing frequency scanning is actually a minimum requirement for all wireless technologies;

  • discovery and update of service capabilities (SCs) of CARs. In general, an SC is the set of parameters (available bandwidth, price, security support, radio access technology, etc.) that characterizes the network service provided to the MN. An SC includes all the most important parameters that influence the handover decisions. Thus, the final aim is to drive the MN operation according to specific objectives, such as (i) to balance the traffic load on the access network, (ii) to drive the MN towards the cheapest wireless access, (iii) to drive the MN towards an access with enough bandwidth to maintain the same level of QoS.

A CARD protocol entity runs at each AR. It has to collect information concerning the surrounding wireless coverage map. This information is the address mapping and service capabilities of every AR which has part of its coverage area superimposed by that of its neighboring AR. Starting from layer 2 information from an MN, the outcome of the protocol is the set of CARs for that MN, together with the relevant SCs. This set of information is used by the TAR algorithm to select the “best” AR, on the basis of a specific metric. Clearly, this mechanism can be used in the case of a wireless coverage, which offers many options for selecting an AR. This assumption will be referred to below as dense wireless coverage. This assumption is realistic for an operator offering a seamless wireless network service, which can deploy a CARD mechanism to support TAR.

A CARD procedure is typically network-assisted. No signaling is exchanged between the MN and any of the CARs before handing over. The procedure runs in the background within the network while the MN is connected to any AR. It is worth noting that, if mobile terminals have multiple network cards and are able to contemporarily manage multiple IP connections, then the MN may be directly informed by CARs, without any assistance from the network. However, this is not a typical scenario and it is not taken into account in this paper. We assume only that MNs can listen to the L2 beacons transmitted by APs which cover the MN position.

We remark that, in principle, a CARD solution can work in a heterogeneous wireless access scenario and can assist inter-technology handovers, if mobile terminals are multimode.

The aim of this paper is threefold.

  • 1.

    Our first goal is to describe and analyze the CARD approaches defined by the IETF Seamoby WG, which has inspired the research in this field. They present both a distributed solution (handover-based solution) and a server-based solution [2] to build the wireless coverage map at ARs (discovery phase). The basic idea of the handover-based solution is that two ARs are discovered to be neighboring after a plain handover of an MN has occurred between them. As regards the server-based approach, the idea is that each AR must register with a centralized server by providing its own IP address, together with all the L2 IDs of its APs. In both cases, the SCs update mechanism (steady phase) is pull-mode oriented.

  • 2.

    The second goal of this paper is to present a novel CARD approach. We assume that the network is managed by a single operator (Connection Service Provider, CSP), which provides IP connectivity. The result of this assumption is that without any commercial agreement, there are no logical reasons for a CSP to induce its mobile customer to move to other access domains, since this would imply a loss of traffic and revenues in favor of competitors. In addition, any CSP is typically unwilling to give other organizations, such as competitor CSPs, confidential information regarding its own network (current bandwidth, security policies, etc.). Thus, our CARD procedure is assumed to be confined within a single administrative domain. The proposed push-mode-multicast CARD (PMM CARD) is fully distributed and allows each AR to dynamically self-construct a map of the surrounding wireless coverage. The PMM CARD is distributed; this avoids the classic problems associated with centralized solutions (single point of failure, scalability, performance bottleneck, denial of service attacks to centralized servers …). The main innovation of our approach with respect to the Seamoby proposal [2] is the use of push-mode multicast transmissions. Two levels of multicast are defined. The former is a global, domain-wide, high level multicast group. It is used to implement address translation functions during the discovery phase. The latter consists of local, AR-relevant, low-level multicast groups. They are used for updating service capabilities during the steady phase. Push-mode multicast transmissions are used in order to both highly reduce the latency due to explicit queries to a remote entity and efficiently distribute CARD information.

  • 3.

    The final step of this paper is the performance evaluation of the PMM CARD approach and the comparison with the other IETF solutions in a homogeneous 802.11b access network. To this end, we have developed a theoretical analysis to evaluate the signaling burden (i.e., the cost of CARD in terms of network resources) of the different CARD solutions. In addition, we have adapted the NS-2 simulator [17] to evaluate the time needed to complete the wireless coverage map at each AR (discovery time) for the PMM and IETF solutions. Results show the effectiveness of the PMM solution.

This paper is structured as follows. In the next section, we summarize the state of the art about CARD. In Section 3, our PMM CARD procedure is described in detail, together with some considerations on security aspects. In Section 4, we present the approach used to analyze and compare the performance of both the PMM CARD procedure and the IETF solutions, in terms of signaling burden. The corresponding numerical results are shown in Section 5, along with the simulation results regarding the discovery time. Section 6 includes an overall critical analysis of the different CARD solutions. Finally, Section 7 reports some concluding remarks.

Section snippets

Related work

In order to enable seamless IP-layer handover, a CARD solution must cover two important tasks:

  • (i)

    the reverse address translation from L2 IDs;

  • (ii)

    the discovery and update of SCs.

The former function is needed to speed-up the handover process. An AP is candidate if the MN listens to its L2 beacons during frequency scanning. Through reverse address translation, MN is quickly informed of the IP address of the AR to hand over to, thus it does not need to wait for the MIP router advertisement and may execute

The PMM CARD

In this section, we present the push-mode-multicast CARD (PMM CARD), which is a proposal for performing the CARD procedure within an administrative domain.

The PMM CARD is network-assisted, distributed and based on push-mode multicast transmission. As regards the discovery phase, we follow a distributed L2 beacon-based approach. It allows us:

  • to avoid the drawbacks of a centralized solution,

  • to avoid the bootstrap handover,

  • to improve the precision and accuracy of the coverage information,

  • to

Signaling load evaluation

In this section, we evaluate the signaling burden of both the PMM CARD and IETF solutions.

To this end, we analyze the discovery phase (construction of CARD tables) and the steady phase (SC update) separately.

The analysis does not take into account the amount of signaling exchanged on the radio interface. The rationale of this choice is that the CARD signaling load is generally more significant in the core network, and in the wireless segment it is nearly the same for each solution, and

Numerical results

In this section, we show some numerical results relevant to the PMM and the IETF CARD solutions, expressed in terms of discovery time and signaling burden. The discovery time is estimated through a simulation campaign. The signaling overhead is obtained by the results of the theoretical analysis shown in the previous section.

Comparison of CARD solutions

In conclusion, Table 4 reports a comparison of the main peculiarities of the different CARD approaches.

Note that all CARD/TAR mechanisms need a mobile terminal to be able to listen to the L2 beacons and to communicate them to the network side. Thus, CARD/TAR functions need to be implemented on both the terminals and ARs. This feature would also imply a number of business aspects, such as: (i) has the TAR selection to be triggered by the user or by the operator? (ii) who is in charge of

Conclusion and future work

The framework of this paper is an administrative domain which provides advanced network support in a heterogeneous wireless scenario. In this scenario, we propose an intra-domain, distributed CARD procedure (named PMM CARD), able to dynamically self-construct a map at each AR of the surrounding wireless coverage, and to discover and update the service capabilities. The main novelty of our proposal is the use of push-mode multicast transmissions. This approach enables some aspects of the

Acknowledgments

The authors would like to thank the anonymous referees for their suggestions and comments that have improved the quality of the paper.

This work was performed under co-financing by the Italian Ministry of Education, University, and Research (MIUR) within the FIRB project PRIMO.

Dario Di Sorte currently works at the Dipartimento di Ingegneria Elettronica e dell’Informazione (DIEI) of the University of Perugia. He received the Ph.D. degree from the University of Florence in June 2003, and the “Laurea” degree in Electronic Engineering magna cum laude from the University of Perugia in January 2000.

His research interests focus on a number of issues in IP networks: quality of service; pricing; content delivery networks; candidate access router discovery.

He has been involved

References (23)

  • J. Manner, M. Kojo, Mobility related terminology, IETF RFC 3753, June...
  • M. Liebsch et al., Candidate access router discovery, IETF RFC 4066, July...
  • J. Loughney, M. Nakhjiri, C. Perkins, R. Koodli, Context transfer protocol (CXTP), IETF RFC 4067, July...
  • J. Kempf, Problem description: reasons for performing context transfers between nodes in an IP access network, IETF RFC...
  • C. Perkins, IP mobility support for IPv4, IETF RFC 3344, August...
  • D.B. Johnson, C.E. Perkins, Mobility support in IPv6, MobiCom’96, Rye, NY, USA, November...
  • A.T. Campbell

    Comparison of IP micromobility protocols

    IEEE Wireless Communications

    (2002)
  • D. Funato et al., Geographically adjacent access router discovery protocol, Internet draft,...
  • D. Trossen, G. Krishnamurthi, H. Chaskar, R.C. Chalmers, E. Shim, A dynamic protocol for candidate access-router...
  • R. Koodli, Fast handovers for mobile IPv6, IETF RFC 4068, July...
  • K. El Malki, Low latency handoffs in mobile IPv4, Internet draft, draft-ietf-mobileip-lowlatency-handoffs-v4-09.txt,...
  • Cited by (0)

    Dario Di Sorte currently works at the Dipartimento di Ingegneria Elettronica e dell’Informazione (DIEI) of the University of Perugia. He received the Ph.D. degree from the University of Florence in June 2003, and the “Laurea” degree in Electronic Engineering magna cum laude from the University of Perugia in January 2000.

    His research interests focus on a number of issues in IP networks: quality of service; pricing; content delivery networks; candidate access router discovery.

    He has been involved on the IST projects SUITED, WHYLESS.COM, FIFTH, and SIMPLICITY, sponsored by the European Union, and on the FIRB projects PRIMO and VICOM, sponsored by the Italian Government.

    He is co-author of a number of papers, published in scientific international conferences and journals.

    Mauro Femminella received his “Laurea” degree in Electronic Engineering in 1999, magna cum laude with publication of his thesis, and earned the Ph.D. degree in Electronic Engineering in 2003, both at the University of Perugia, Italy. He was Consulting Engineer for the University of Perugia, and for the consortia CoRiTel and RadioLabs. Actually he holds a position as contract researcher at the Department of Information and Electronic Engineering at the University of Perugia.

    He was involved in a number of research projects co-funded by the European Union (programs ACTS and IST), by the Italian Ministry for Education, Higher Education and Research (MIUR), and by the European Space Agency (ESA).

    He is co-author of a number of papers in international conferences and journals.

    His research interests focus on design and performance evaluation of satellite networks, content delivery networks, IP quality of service and IP mobility.

    Piacentini Leonardo received his “Laurea” degree in Electronic Engineering (curriculum Telecommunications) magna cum laude from the University of Perugia in January 2001.

    Since January 2002 he has been working as a Ph.D. Student of the Department of Electronic and Information Engineering (DIEI) of the University of Perugia.

    His actual research is focusing on mobility in wireless networks.

    After graduated, he worked on a MIUR project named RAMON (Reconfigurable Access module for Mobile computiNg applications).

    Gianluca Reali has been a researcher of the Department of Information and Electronic Engineering of the University of Perugia since 1997. He received the “Laurea” degree in Electronic Engineering from the University of Perugia in 1991. Then he collaborated with Alenia Spazio S.p.A. working on VSAT networks. He has also acted as Consulting Engineer for the Institute of Electronics of the University of Perugia and CRA in Rome. He received his Ph.D. degree in Telecommunications from the University of Perugia in 1997, working on Spread Spectrum techniques and CDMA. His present research activity is in IP QoS techniques, particularly in switching, transport, and network management protocols. He has been involved in the European ACTS projects CABSINET and ASSET, and in the European IST projects SUITED, WHYLESS.COM, and SIMPLICITY.

    View full text