The Pre-constructing Data Path

In our research, we implement a pre-constructing method of data path establishment. In this mobility method, we assume that the mobility path of MS is known, so the target BS in the predictive mobility path is a candidate BS for data path

establishment. The pre-constructing data path method knows where the MS going, and then pre-establishes the data path between the BS and the ASN-GW. Even the roaming of MS is inter-ASN mobility, we can pre-constructing the R4 and R6 data path. When MS initial entries the network and establishes the data path among anchor ASN-GW and serving BS, then the anchor ASN-GW will pre-constructing the data path in the predictive data path. The pre-constructing procedure uses the PATH_REG_REQ/PATH_REG_RSP/PATH_REG_ACK messages for data path establishment but the pre-constructing data path is not active when MS does not yet roam to the candidate BS. When MS is roaming to the candidate BS (target BS), MS will send the PATH_REG_REQ/PATH_REG_RSP messages to make the pre-constructing data path active. In this method, the message flow is similar to the other methods above. The only different is that when MS initial entries, the data path establishes predictive, and when MS roams, the only thing to do about the data path is registering it active. MS does not establish the data path when roam because it is pre-constructing. Figure 10 shows the predictive data path establishment of MS. We can see that the green line of R6 is current data path when MS connect to serving BS, and the blue line of R4 and R6 is that the anchor ASN-GW according to the predictive path of MS to pre-establish the data path.

Next chapter we will implement the inter-ASN mobility by R4 mobility and CSN anchored mobility. Also, we implement the predictive data path establishment by using the R4 data path pre-constructing. In the Final, we evaluate and compare these methods in data packet loss.

24 

Figure 10 The Predictive Data Path Establishment

Chapter 4

Performance Evaluation

In this chapter, we implement the mobility methods in chapter 3 on the WiMAX ASN-GW and BS emulator and evaluate the packet loss of handoff.

4.1 Evaluation Tool

The Institute for Information Industry (III) [12] is in a leadership of WiMAX in Taiwan and the M-Taiwan project that III involved is a goal to make Taiwan be a completely mobility environment. The III is now researching and designing the WiMAX BS and the BS developed by III has already passed the testing by connecting and interflowing with the WiChorus [13] WiMAX ASN-GW. We implement the mobility methods in NWG End-to-End Network Architecture by the ASN-GW and BS emulator supported by III, and the MS is also an emulator that developed in the BS. Although the ASN-GW and BS are just emulators, if we port them into hardware with WiMAX physical specification, the BS and ASN-GW can be real work devices of WiMAX.

26 

4.2 Evaluation Environment

Figure 11 Evaluation Architecture

Figure 11 is the evaluation architecture of WiMAX End-to-End Network Architecture in our article. There are two subnets, ASN-GW_1 connects to the BS_1 and BS_2 is an independent ASN network and also as ASN-GW_2 with BS_3 and BS_4. The whole network is under the local area network (LAN) that has private IP address. The corresponding node (CN) is a packet generator sending packet to MS and the home agent (HA/MIP HA) will receive the packet and repack it in IP-in-IP header then send to the foreign agent (FA/ASN-GW). Following Table 1 shows the experiment parameters.

Data Packet size 64 bytes/packet

Protocol UDP

Packet Sending interval Every 60 millisecond/0.06 second BS to BS interval 1 second

Experiment duration 100 seconds Handoff scenario Hard handoff

Table 1 The Setting of Experiment

According to the architecture of Figure 11, the MS will move from BS_1 to BS_4 in 100 seconds. At the movement, handoff occurs three times in BS_1 to BS_2, BS_2 to BS_3, and BS_3 to BS_4. When MS leaves the serving BS, it will arrive at the target BS in next second, so the interval between BS to BS is 1 second. Between the MS roaming interval the CN will constantly sending packet to the MS, and we will count the packet loss of MS.

4.3 Evaluation Cases

In this experiment, there are three kinds of handoff will occur. First is the ASN anchored mobility (R6 mobility) between BS_1 and BS_2, BS_3 and BS_4. Between BS_2 and BS_3, the inter-ASN mobility, we will use the CSN anchored mobility (R3 mobility) in case 1 to compare with the R4 mobility in case 2. The CSN anchored mobility will register the new ASN-GW/FA of MS to the HA and the packet will redirect to the new ASN-GW/FA, and the R4 mobility will not change the anchor ASN-GW but build an data path (GRE tunnel) between anchor ASN and serving ASN that the packet from CN will still send to the anchor ASN-GW then follow the data path to the serving ASN-GW and the serving BS. In the case 3, we use the pre-constructing data path in this experiment. We assume that the anchor ASN-GW is already know the moving path of MS and pre-establishes the data path for BS_2,

28 

BS_3, and BS_4, the BS_3 data path establishment uses the R4 mobility data path concept. The pre-established data path will be active when MS roam to the target BS.

4.4 Numerical Results

Figure 12 and Figure 13 shows the evaluation result of the two handoff mechanisms, CSN anchored mobility (case 1) and R4 mobility (case 2). We can see the packet sequence number accumulates in Figure 12 and Figure 13, there are three broken points in the graph, the first and the third broken points are resulting in the ASN anchored mobility between BS_1 and BS_2, BS_3 and BS_4. The second broken point is the handoff between BS_2 and BS_4, and we can clearly see that in Figure 12 the broken interval is longer than that in Figure 13. The Result shows that if we could use the R4 mobility to replace the R3 mobility in roaming between different ASN domains, the packet loss will decrease. The pre-constructing data path method (case 3) evaluation result is showed in Figure 14, there are also three broken point in the data packet accumulation graph. We could saw that pre-constructing data path case has fewer packet losses than the other two cases.

Table 2 shows the total packet loss of handoff in the three cases, we could find that the R4 mobility handoff could decrease the 20% packet loss than the R3 mobility handoff. But the pre-constructing case does not show the significant improvement than the R4 mobility method, although it still has less 10% packet loss. We can figure out the result is that the emulator is not in a real WiMAX hardware and the procedure delay occurs because the experiment setting is six emulator in one kernel, and the pre-constructing data path is still need the PATH_REQ/PATH_RSP message to make the data path active. Although the active message is shorter and fewer, the procedure also speed time in handoff and it causes the packet loss. Table 2 also shows the data packet delay in case 1 and case 2, it apparently presents that the R4 mobility will

spend more time sending a packet than R3 mobility because R4 mobility will not route the path to the serving ASN-GW. Even the data packet delay is a tinny difference, but the experiment is on the emulator. If we experiment in the real WiMAX environment, the data packet delay will increase and R4 mobility also has its trade-off in MS roaming. We will test this experiment in real WiMAX hardware in future, and we believe that the pre- constructing data path method will show better significant result than other methods.

Figure 12 Case 1: CSN Anchored Mobility (R3 mobility)

30 

Figure 13 Case 2: R4 mobility

Figure 14 Case 3: Pre-establish Data Path

Case 1:

R3 mobility

Case 2:

R4 mobility

Case 3:

Predictive

R6 Mobility between BS_1 and BS_2 59 42 35

Inter-ASN mobility between BS_2 and BS_3 74 45 40

R6 Mobility between BS_3 and BS_4 42 50 38

Total packet loss (loss/total) 175/1640 137/1640 113

Packet loss rate 10.67% 8.35% 6.89%

The inter-ASN handoff delay 4.44 sec 2.7 sec 2.4 sec The data packet delay after inter-ASN HO 64ms 93ms

Table 2 The Packet Loss Comparison While Handoff

Chapter 5

Conclusion and Future Work

In our thesis, we implement the R4 mobility between two ASN-GWs to replace the CSN anchored mobility of inter-ASN mobility and use the R4 mobility to implement a pre-establishment data path method. In the highly mobility environment, the seamless handoff is an very popular topic in WiMAX network, and the improvement of handoff could not only in the IEEE 802.16 PHY and MAC layer protocol but also could be in the End-to-End Network Architecture. Because the network architecture is the special feature of WiMAX, the network environment assistant handoff would be a workable method to improve the handoff in WiMAX network. We have emulated the WiMAX End-to-End Network Architecture and evaluated the packet loss of R3, R4, and R6 mobility in this article, but it only uses hard HO method between MS and BS.

In the future, we will do the experiment in the real WiMAX hardware and try other mobility methods according to the network architecture or the MS/BS handoff scenarios to improve the network assistant handoff by the design of ASN-GW and BS.

And also we will implement the data integrity method in ASN-GW for buffering the

32 

data packet to reduce the data packet loss when MS roaming. Because ASN-GW is an central role in ASN network, the information gathering in ASN-GW can be numerous MSs, so how to make the resource in better using is also a research topic in our future work.

Bibliography

[1] WiMAX Forum. [Online] http://www.wimaxforum.org

[2] IEEE Std 802.16-2004 ^TM, IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interface for Fixed Broadband Wireless Access Systems, Oct 2004.

[3] IEEE Std 802.16e-2005^TM, IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems, Feb 2006.

[4] IEEE Std 802.16g-2007 ^TM, IEEE Standard for Local and Metropolitan Area Networks - Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems - Management Plane Procedures and Services, Dec 2007.

[5] WiMAX Forum Proprietary, “WiMAX End-to-End System Architecture (Stage 2:

Architecture Tents, Reference Model and Reference Points, Stage3: Detailed Protocols and Procedures),” January 11, 2008 Release 1.2.1.

[6] D. Farinacci, T. Li, S. Hanks, D. Meyer, P. Traina, "Generic Routing Encapsulation (GRE)," RFC 2784, Internet Engineering Task Force, March 2000.

[7] C. Perkins, Ed., “IP Mobility Support for IPv4,” RFC 3344, Internet Engineering Task Force, August 2002.

34 

[8] D. Johnson, C. Perkins, J. Arkko, “Mobility Support in IPv6,” RFC 3775, Internet Engineering Task Force, June 2004

[9] D Johnston, J Walker, "Mutual Authorization for PKMv2," IEEE C802.16e-04/229, 2004.

[10] C Rigney, A Rubens, W Simpson, S Willens, “Remote Authentication Dial In User Service (RADIUS),” RFC 2865, Internet Engineering Task Force, June 2002.

[11] K Leung, G Dommety, P Yegani, K Chowdhury, “WiMAX Forum/3GPP2 Proxy Mobile IPv4,” draft-leung-mip4-proxy-mode-07, Internet Engineering Task Force, Feb, 2008.

[12] Institute for Information Industry (III), organization in Taiwan. [Online]

http://www.iii.org.tw/

[13] WiChorus. [Online] http://www.wichorus.com/

[14] P. lyer, N. Natatajan, M. Venkatachalam, A. Bedekar, E. Gonen, K. Etemad, P.

Taaghol, “All-IP Network Architecture for Mobile WiMAX,” Mobile WiMAX Symposium, 2007. IEEE

[15] S. Das, T. Klein, A. Rajkumar, S. Rangarajan, M. Turber, H. Viswanathan,

“System Aspects and Handover Management for IEEE 802.16e,” Bell Labs Technical Journal, 2006.

[16] M. Ergen, “The Access Service Network in WiMAX: The Role of ASN-GW,”

WiChorus, Inc.