Distributed dynamic mobile multicast☆
Research highlights
► We propose an analytical model to formulate the cost of multicast tree reconfiguration and multicast packet delivery. ► We apply a Markov chain to analyze the mobility of MNs in a 2D mesh network. To reduce the computation complexity, we aggregate the Markov states by leveraging the mobility symmetry. ► An iterative algorithm is derived to quickly find the optimal service range that balances the cost of multicast tree reconfiguration and multicast packet delivery.
Introduction
The rapid progress of mobile networks has led to the tremendous demand for mobile services. One such service is mobile multicast because it outperforms the basic broadcast strategy by sharing resources along common links, while sending messages to a set of predefined destinations [22]. Due to its efficiency and flexibility, mobile multicast has gained a wide spectrum of applications, such as video-conferencing, video-on-demand, stock-quote distribution, and so on. Therefore, mobile multicast has received significant attention in recent years.
However, mobile multicast poses significant challenges because it must deal with not only dynamic group membership, but also dynamic locations of its members. To this end, Mobile IP proposes two basic schemes: the remote subscription (RS) and bi-directional tunneling (BT) [17], [9].
In the RS, a mobile node (MN) is required to re-subscribe to its desired multicast groups every time it enters a new subnet. Thus, the multicast packets will be directly forwarded to the MN. However, the frequency of multicast tree reconfiguration is higher because it is directly proportional to users’ handoff frequency.
In the BT, on the other hand, the home agent (HA) joins in the desired multicast groups instead of the MNs. Hence, all multicast packets and signals are sent or received through the HA. Since this scheme hides a user’s mobility from other members of the group [12], no multicast tree needs to be updated after handoffs. However, the BT introduces the triangle routing problem, which makes the multicast packet delivery path far from optimal.
According to the above two mechanisms, RS has a high multicast tree reconfiguration cost while a low multicast packet delivery cost, which is exactly the opposite of the scenario in the BT. It is noteworthy that both multicast tree reconfiguration and multicast packet delivery costs are critical to the performance of mobile multicast. This is because a higher multicast tree reconfiguration cost incurs excessive signaling redundancy (from the viewpoint of network) and multicast service disruption time (from the viewpoint of users). This situation is worse in wireless mesh networks (WMNs) because the scarce and possibly asymmetrical wireless bandwidth require the amount of control signaling to be limited. In contrast, a higher packet delivery cost results from a longer packet delivery path and hence time. In multi-hop wireless networks, the longer the packet delivery path, the worse is the quality of service an MN receives.
Thus, a trade-off between these two costs is very important. This problem is similar to the trade-off between registration and packet delivery delays in mobile unicast. Hence, a natural idea to solve the trade-off problem in mobile multicast is to learn the idea of regional mobility management from mobile unicast schemes [20], [4], [13], [24], thus leading to what are called the region-based mobile multicast schemes [12], [21], [23], [29].
The main idea of region-based mobile multicast schemes is to divide the whole network into multiple service ranges and deploy a mobile multicast agent (MMA) within each service range. This task requires that the MMAs must join in the desired multicast group and forward multicast packets to users within the service range by tunneling or multicasting. In region-based mobile multicast schemes, multicast packet delivery path is closer to the shortest because the multicast packets need not go through the HA before being forwarded to the remote mobile receivers. In addition, the frequency of multicast service disruption times is reduced because the multicast tree is reconstructed only when the MN moves out of the service region.
However, most region-based mobile multicast schemes fail to discuss how to determine the size of the service range, which is critical to network performance. We use Fig. 1 to illustrate this issue, where the MMA is denoted by the circle with shadow. In this paper, we define the size of the service range as the set of foreign agents (FAs) whose distances to the MMA are within the given hops. According to this definition, the scope surrounded by the dash line denotes the MMA’s service range. As shown in Fig. 1(a), if the service range is larger, the MMA handoff probability is smaller resulting in lower multicast tree reconfiguration probability. However, a larger service range extends packet delivery time, thus decreasing throughput especially in the case of a multiple wireless nodes scenario due to their interference [10].
If the service range is smaller, the multicast packets may be delivered along a shorter path. However, as shown in Fig. 1(b), the smaller the service range, the more frequent is the MMA handoff, which leads to frequent multicast tree reconfiguration.
Since finding an optimal service range can minimize the overall time for multicast packet delivery and multicast service disruption, we propose a distributed dynamic mobile multicast scheme to dynamically determine the optimal service range for an MMA according to the mobility and service characteristics of a user. In particular, we make the following contributions:
- (i)
We propose an analytical model to formulate the cost of multicast tree reconfiguration and multicast packet delivery as functions of the service range.
- (ii)
To establish the analytical model, we apply a Markov chain to analyze the mobility of MNs in a 2D mesh network. Because the complexity of computing steady probability is high, we aggregate the Markov states by leveraging the mobility symmetry. In addition, a method is proposed to obtain the aggregate state set and the transition probability between any two aggregate states.
- (iii)
An iterative algorithm is derived to quickly find the optimal service range that balances the cost of multicast tree reconfiguration and multicast packet delivery. However, the computation of the optimal service range is an overhead of . To reduce this overhead, we present a mobile multicast architecture which groups the MMAs whose users have similar mobility and service characteristics into the same range region (SRR). Thus, only one MMA needs to compute the optimal service range for other MMAs in the same SRR.
is also compatible with any multicast tree configuration algorithm. The MMA of an MN is decided by not only the MN’s location, but also the mobility and service characteristics of the MHA’s users. This makes each FA act as either a regular FA or an MMA. In other words, network traffic is feasibly allocated to each FA. Simulation experiments demonstrate that the optimal service range enhances network performance.
This paper is organized as follows: Section 2 presents related work. The overview of is given in Section 3, while Section 4 presents the problem formulation and modeling. Section 5 discusses how to obtain the optimal service range. Section 6 describes some implementation issues and Section 7 presents the performance evaluation. Finally, we conclude the paper in Section 8.
Section snippets
Related work
In RS, an MN needs to re-subscribe to the FA whenever a handoff occurs. In this scenario, the packet loss rate might vary between 1% and 30% [18]. To reduce the packet loss, Jiunn-Ru and Wanjiun [8] proposes a mechanism permitting an old agent to deliver the packets received by an MN while roaming into a new one. A similar method is proposed in [11], which makes an MN receive the multicast packets soon after handoff through the tunnel between the new FA and the old one. Jiannong et al. [7]
Overview of
Like region-based mobile multicast schemes, introduces MMAs as the entities for managing mobile multicast within the service ranges. The aim of is to find the optimal service range for each MMA to realize the trade-off between the costs for multicast tree reconfiguration and multicast packet delivery. Before proceeding further, let us define a few terminologies.
Problem formulation and modeling
As described above, is the key to . To compute , we must derive the overall cost beforehand. In light of Eq. (2), is a function of . In the following, we first assume that the MMA region is confined by , and then we find the value of that minimizes to obtain .
By Definition 1, includes and , where is the average cost for delivering multicast packets from the MMA to the FA and is the average cost for multicast tree configuration. Assuming that the
Solution of optimal service range
According to Definition (1), is the value of that minimizes the cost function. Since can only be an integer and the cost is not a continuous function of , the following method is adopted to derive [25]. First, we define the following functions:
Then formulas (24), (25) lead to the following equation:
According to the definition of the minimizing function described in
Implementation issues
For adapting well to the dynamics of mobile networks, is calculated based on the user mobility and service characteristics, where the mobility characteristic is captured by the average FA residence time (i.e., ), while the service characteristic is captured by the average multicast packet arrival rate (i.e., ). These parameters may change dynamically, implying that can be different and adjusted from time to time. Now can be calculated by the method introduced in [1], while the
Performance analysis
This section compares the performance of with RS, BT and RBMoM. Among these schemes, RBMoM and have limited service ranges, so they can be grouped as the service-range-based mobile multicast (SRBMM) scheme. In fact, both RS and BT are the extremes of SRBMM [12]. If the service range , SRBMM is the same as BT; If the service range = 0, SRBMM is the same as RS. In this section, we first perform simulation experiments to observe the change of tunnel distance as well as the number of
Conclusion
In an MMA only serves users, whose FAs are within the optimal service range. The optimal service range can realize the trade-off between multicast service disruption time and multicast packet delivery time. This trait makes effectively adapt to the dynamics of mobile networks. For finding the optimal service range, we model the MN’s movement in a 2D mesh network using Markov chain. To reduce the computational complexity, we aggregate the states by using the mobile movement symmetry.
Acknowledgments
The authors are grateful to the anonymous reviewers for constructive comments which helped to improve the quality of the manuscript.
Yong Cui, born in 1976, received the Ph.D. degree in 2004 from Tsinghua University, PR China. He is now an associate professor at the Department of Computer Science in Tsinghua University. He directed several national R&D projects, including some projects funded by the Chinese Next Generation Internet Plan (CNGI), a project funded by the National Natural Science Foundation of China (NSFC), and a project funded by the Development Plan of the State High Technology Research of China (863). He also
References (31)
- et al.
Supporting IP multicast for mobile hosts
ACM/Springer Mobile Networks and Applications
(2001) - Y. Chen, M. Huang, A novel MAP selection scheme by using abstraction node in hierarchical MIPv6, in: Proc. of IEEE...
- et al.
A 2-D random-walk mobility model for location-management studies in wireless networks
IEEE Transactions on Vehicular Technology
(2004) - et al.
mMoM: efficient mobile multicast support based on the mobility of mobile hosts
IEICE Transactions on Communications
(2002) - et al.
TeleMIP: telecommunications-enhanced mobile IP architecture for fast intra-domain mobility
IEEE Personal Communications
(2000) - et al.
Mobile multicast (MoM) protocol: multicast support for mobile hosts
- C. Jelger, T. Noel, Supporting mobile SSM sources, in: Proc. of IEEE Global Telecommunications Conference, 2002, pp....
- C. Jiannong, S. Mullai, D. Leung, C. Hui, An efficient multicast service switching protocol in mobile IP, in: Proc. of...
- et al.
Mobile multicast with routing optimization for recipient mobility
IEEE Transactions on Consumer Electronics
(2001) - D. Johnson, C. Perkins, J. Arkko, Mobility Support in IPv6. RFC3775,...
IDMP-based fast handovers and paging in IP-based 4G mobile networks
IEEE Communications Magazine
A reliable advanced-join system for data multicasting in ITS networks
IEEE Transactions on Intelligent Transportation Systems
Cited by (2)
Optimized routing algorithm for mobile multicast source in wireless mesh networks
2016, Proceedings - 2015 11th International Conference on Innovations in Information Technology, IIT 2015A new high performance approach: Merging optimal multicast sessions for supporting multisource routing
2013, Journal of Supercomputing
Yong Cui, born in 1976, received the Ph.D. degree in 2004 from Tsinghua University, PR China. He is now an associate professor at the Department of Computer Science in Tsinghua University. He directed several national R&D projects, including some projects funded by the Chinese Next Generation Internet Plan (CNGI), a project funded by the National Natural Science Foundation of China (NSFC), and a project funded by the Development Plan of the State High Technology Research of China (863). He also participated as a major player in a project funded by the National Basic Research Plan of China (973 Program) and a Key Project of the National Natural Science Foundation of China. Having published more than 60 papers in international journals and conferences, he also holds more than 20 Chinese patents. His major research interests include wireless mesh networks, mobility management, the next generation Internet architecture, quality of service and distributed routing protocols.
Shengling Wang, born in 1978, received the Ph.D. degree in Computer Science in 2008 from Xi’an Jiaotong University. After that, she worked at Tsinghua University as a postdoctoral fellow. Now, she is an assistant research fellow at the Institute of Computing Technology, Chinese Academy of Sciences. Her major research interests include QoS, wireless/mobile networks, internet of things.
Sajal K. Das, born in 1976, received the Ph.D. degree in Computer Science in 1988 from University of Central Florida, Orlando. He is now a Professor of the Department of Computer Science and Engineering in University Texas at Arlington. He is director of Center for Research in Wireless Mobility and Networking (CReWMaN). His major research interests include cellular mobile communication and computing, wireless multimedia, sensor networks/computer security, pervasive computing, broadband networking, parallel and distributed processing, parallel algorithms and data structures, parallel discrete-event simulation and multiprocessor interconnection networks.
- ☆
This work is supported by the National Natural Science Foundation of China (No. 61003225, No. 60911130511, No. 60873252) and the National Major Basic Research Program of China (No. 2011CB302702).