Transcript Slide 1
Smartening the Environment using Wireless Sensor Networks in a Developing Country A TEST-BED ANALYSIS FOR SEAMLESS MIPV6 HANDOVER IN HETEROGENEOUS ENVIRONMENT Mohammad Moshee Uddin, International Islamic University Malaysia Al-Sakib Khan Pathan, International Islamic University Malaysia Shariq Haseeb, MIMOS Berhad, Kuala Lumpur, Malaysia Mohiuddin Ahmed, Jazan University, Saudi Arabia Presenter Al-Sakib Khan Pathan Department of Computer Science International Islamic University Malaysia, Malaysia Outline of the Presentation • • • • • • Introduction Background Objective Experimental Setup Analysis and Results Concluding Remarks ISCE, June 14-17, 2011, Singapore 2 Introduction • Use of multiple network interfaces is becoming more common with a mobile node (MN). • Now-a-days, almost every hand-held device has multiple network interfaces built-in – – – – – Wi-Fi Ethernet WiMAX Bluetooth UMTS • Multiple network interfaces converged network – Ubiquitous communications ISCE, June 14-17, 2011, Singapore 3 Introduction (Continued) • Mobile IPv4 (MIPv4) has become a part of the solution for the mobility support system to have ubiquitous communication. • Yet, it could not solve lots of problems because of its limitations to support wide-scale applications such as Peer-to-Peer (P2P) applications, addressing limitations as well as IPsec (Internet Protocol Security), etc. ISCE, June 14-17, 2011, Singapore 4 Background: MIPv6 • MIPv6 is a key protocol which allows a node to have ubiquitous communication with the help of mobility support system. • It allows MN to change its point of attachment without changing the “Home Address” of MN. So any packet may still be routed regardless of any point of attachment as long as it is attached to the Internet. ISCE, June 14-17, 2011, Singapore 5 MIPv6 (Continued) • Furthermore, multiple network interfaces also can be handled by MIPv6 protocol to support heterogeneous mobility. • For instance, the movement from one Wi-Fi segment to Ethernet segment or Ethernet segment to Wi-Fi or Wi-Fi segment to WiMAX, and so forth, if the “Home Address” remains the same. ISCE, June 14-17, 2011, Singapore 6 Main Objective of the Work • Test-bed experimentations of vertical MIPv6 handover performance (i.e., from Ethernet segment to Wi-Fi segment and vice versa) to evaluate – Handoff latencies – Packet losses while multiple interfaces are simultaneously associated with different types of networks. ISCE, June 14-17, 2011, Singapore 7 MIPv6 Handover Process ISCE, June 14-17, 2011, Singapore 8 Our Experimental Setup • The test-bed has been implemented with two scenarios: – Scenario1: When MN moves around between home link and foreign link (Figure 1) – Scenario 2: Again when MN moves around foreign links (Figure 2) ISCE, June 14-17, 2011, Singapore 9 Scenario 1 (Figure 1) • • • • • Two PC-based routers: HA and FR One switch One IEEE 802.11abg Access Point One notebook as MN with one Ethernet and one wireless interface built-in One PC-based Correspondent Node (CN). ISCE, June 14-17, 2011, Singapore 10 Scenario 2 (Figure 2) • • • • • Three PC-based routers: HA, FR1, and FR2 Two switches One IEEE 802.11abg Access Point One notebook as MN with one Ethernet and wireless interface built-in One PC-based CN ISCE, June 14-17, 2011, Singapore 11 Experiment Methodology • Figure 1 shows MN is associated with HL via Ethernet & at the same time Wi-Fi network coverage is present. • At this moment, MN is communicating with CN via home link. Since Wi-Fi network is available, MN is also preassociated with Wi-Fi network (a.k.a FR). ISCE, June 14-17, 2011, Singapore 12 Experiment Methodology (Contd.) • At any given time, to create a vertical handover, Ethernet connection has been disconnected manually from the home link. • Then the Wi-Fi interface immediately takes over the data communication from Ethernet. This type of vertical handover has happened from home network to foreign network. ISCE, June 14-17, 2011, Singapore 13 Experiment Methodology (Contd.) • Likewise, while MN is associated with the Wi-Fi network and communicates with CN, at any given time Ethernet cable has been manually plugged into the MN. • After that, Wi-Fi interface has been disconnected from foreign link and Ethernet immediately takes over the handover procedure. • The packets are captured to determine handoff delays and packet losses while MN moves from home link to foreign link and vice versa with multi-homed MN. ISCE, June 14-17, 2011, Singapore 14 Experiment Methodology (Contd.) • In a similar fashion, handoff delays and packet losses have been captured (see Figure 2), while MN moves from foreign link to a new foreign link using multiple network interfaces. ISCE, June 14-17, 2011, Singapore 15 Methodology and Tools • The handover measurement has been conducted 50 times with the multi-homed MN between 2 and 3 seconds interval of router advertisement. • MIPv6 tester [1] tool has been used to capture the heterogeneous handoff latency where it opens bidirectional TCP/UDP packets between MN and CN over the network. • To capture packet loss, ‘iperf’ [2] tool has been used during heterogeneous handover. ISCE, June 14-17, 2011, Singapore 16 Mathematical Analysis • List of Notations LTotal L2 L3 LProbe Total handover latency Total layer 2 handover latency Total layer 3 handover latency Layer 2 latency that scans for available AP LAuth Layer 2 latency that performs authentication LRe-assoc Layer 2 latency that performs re-association LRouter Discovery Layer 3 latency that performs IP address configuration LDAD Layer 3 latency that performs uniqueness on the link Layer 3 latency that performs a message regarding location status Layer 3 latency that performs a message confirming location status LBU LBA ISCE, June 14-17, 2011, Singapore 17 Mathematical Analysis • According to [3], [4], [5], and [6], total handover latency in MIPv6 could be mathematically put as follows: LTotal = (L2 + L3) (1) L2 = LProbe + LAuth + LRe-assoc (2) L3 = LRouter Discovery + LDAD + LBU + LBA (3) • Therefore, LTotal = LProbe + LAhth + LRe-assoc + LRouter Discovery + LDAD + LBU + LBA ISCE, June 14-17, 2011, Singapore (4) 18 Analysis • This total handover latency (equation 4) is calculated once the current AP becomes unavailable and MN associates itself with a new AP during the movement (i.e. a scenario of horizontal handover). – Suppose an MN is associated with network X, at the same time it could be under the coverage of network Y. Therefore, it can preassociate with the network Y while network X is still available. But at any time network X may no longer available, then the communication with MN will be handed over to network Y (X and Y could be any type of network such as Ethernet LAN, Wi-Fi network, WiMAX network, UMTS etc). ISCE, June 14-17, 2011, Singapore 19 Results: Handoff Latency TABLE 1. HANDOFF LATENCY DURING THE MOVEMENT OF MN Max Min Average HN to FN (Sec) FN to FN (Sec) FN to HN (Sec) 1.835 0.518 1.229 0 0 0 0 0 0 Figure 3 ISCE, June 14-17, 2011, Singapore 20 Results Analysis • While MN is handed over to FN from HN with multihomed interfaces, the average delay is 1.835s (Table 1), whereas the average handover delay with single interface is 3.677s as in [7] and 3.447s as in [8]. • As for single interface handover process, it has to maintain total layer 2 and layer 3 procedures following equation 4. • But for multi-homed MN, if one of the interfaces is preassociated with foreign network; layer 2 handover delay can be reduced from total delay. Therefore, we can get from the equation 1, LTotal = (L2 + L3) - L2 = L3 ISCE, June 14-17, 2011, Singapore 21 Results Analysis • Observing this experiment, multi-homed MN also reduces the processing time of layer 3 (router discovery and DAD [Duplicate Address Detection]) except binding update (BU) and binding acknowledgement (BA). ISCE, June 14-17, 2011, Singapore 22 Results Analysis • On the other hand, while MN handover is done from FN to another FN or it returns to home network from any foreign network (Figure 3 and Table 1), minimal handover latency occurs that could not be possible to be detected by the MIPv6 tester tool. • As multi-homed MN is simultaneously associated with the networks, only one of the interfaces would be communicating but HA keeps tracking all interfaces. • This immediate handover procedure time is approximately unnoticeable. So from these experiments and equation number 1, it can be derived, LTotal ≈ 0, if multi-homed MN is pre-associated. ISCE, June 14-17, 2011, Singapore 23 Results: Packet Loss TABLE 2. PACKET LOSS DURING THE MOVEMENT OF MN Max Min Average HN to FN (Sec) FN to FN (Sec) FN to HN (Sec) 1.835 0.518 1.229 0 0 0 0 0 0 Figure 4 ISCE, June 14-17, 2011, Singapore 24 Results: Packet Loss • Figure 4 shows the total packet loss in percentages of 50 trials during handoff and Table 2 summarizes maximum, minimum and average packet loss. • With multi-homed MN, average packet loss is about 3.85% while it moves from home network to foreign network. At this point, packet loss occurs during binding update and binding acknowledgement processes at layer 3 as in handoff latency. • Packet loss is proportional to handoff latency. LTotal ∝ PTotal (where PTotal is total packet loss) ISCE, June 14-17, 2011, Singapore 25 Results: Packet Loss • Again, while multi-homed MN moves from one foreign network to another or returns to home network, minimal packet loss may occur that also could not be detected, similar to the handoff latency. Therefore, packet loss is approximately unnoticeable as shown in Figure 4 and Table 2 during the MN’s movement from foreign network to another or its return to Home network. • Hence, PTotal ≈ 0%, if multi-homed MN is pre-associated. ISCE, June 14-17, 2011, Singapore 26 Concluding Remarks • MIPL (Mobile IPv6 for Linux) enables all the features of mobility for heterogeneous environment, yet some delays may occur during handover processes which cause some major packet loss. • There are handoff delays and packet losses during the handover process from HN to FN which degrade the performance of communication. • While MN moves from one FN to another or returns to home network, handoff delay and packet loss are almost unnoticeable and this improves communication process. ISCE, June 14-17, 2011, Singapore 27 Concluding Remarks • Based on our findings we may state that; with the help of MIPv6, a multi-homed MN may perform better vertical handover process while it moves among foreign networks in heterogeneous environment. • As our future work, we would like to perform an extended experiment and analyze other associated parameters to get a detailed understanding of the potential use of the technology for consumer related applications as well as for other application areas. ISCE, June 14-17, 2011, Singapore 28 Major References [1] MIPv6 tester, retrieved June 1st , 2010 from http://www.bullopensource.org/mipv6/tester.php [2] Measure Network Performance with iperf, Retrieved June 1st,2010,from http://www.enterpriseitplanet.com/networking/features/article.php/3659616 [3] D. Johnson, C. Perkins and J. Arkko, “RFC 3775 Mobility Support in IPv6,” URL reference: http://www.ietf.org/rfc/rfc3775.txt (June 2004). [4] M. Siksik, H. Alnuweiri and S. Zahir, “A Detailed Characterization of the Handover Process Using Mobile IPv6 in 802.11 Networks,” IEEE Pacific Rim Conference on Communications, Computers and Signal Processing, Victoria, Canada, August 2005. [5] V. Vassiliou and Z. Zinonos, “An Analysis of the Handover Latency Components in Mobile IPv6,” Journal of Internet Engineering, Vol.3, No.1, December, 2009, pp. 230240. [6] S. Haseeb and G. Kurup, “Performance Analysis of MIPL based Mobile IPv6 Testbed,” Proceedings of the 2007 IEEE International Conference on Telecommunications and Malaysia International Conference on Communications, May 14-17, 2007, Penang, Malaysia. ISCE, June 14-17, 2011, Singapore 29 Major References [7] V. Vassiliou and Z. Zinonos, “An Analysis of the Handover Latency Components in Mobile IPv6,” Journal of Internet Engineering, Vol.3, No.1, December, 2009, pp. 230240. [8] S. Haseeb and G. Kurup, “Performance Analysis of MIPL based Mobile IPv6 Testbed,” Proceedings of the 2007 IEEE International Conference on Telecommunications and Malaysia International Conference on Communications, May 14-17, 2007, Penang Malaysia. ISCE, June 14-17, 2011, Singapore 30 THANK YOU ISCE, June 14-17, 2011, Singapore 31 Questions and Answers [email protected], [email protected] ??? URL: http://staff.iium.edu.my/sakib/ ISCE, June 14-17, 2011, Singapore 32