Sie sind auf Seite 1von 4

2009 IEEE Mobile WiMAX Symposium

On Handover Performance Analysis in Mobile


WiMAX Networks
Yazan A Alqudah, Senior Member, IEEE, Huiqin Yan

Abstract—Mobile WiMAX carries the promise of ubiquitous II. HANDOVER OVERVIEW


broadband wireless access enabling real-time and multimedia
Handover is defined as the process in which an MS
applications. One of the main features for enabling mobility is
Handover. This paper introduces a new methodology using data migrates from the air interface provided by one BS to an air
latency for characterizing handover performance. The new interface provided by another BS.
methodology also enables the detection of issues during handover The preparation for handover starts with the exchange of
by projecting data latency pattern. capability information during the registration process that
pertains to whether the device can support handover, as well
Index Terms — WiMAX, Handover, Latency as other information related to handover configuration. The
MS learns about neighboring BSs through the BS broadcasted
management message: MOB_NBR_ADV. This message
I. INTRODUCTION
contains neighbor information required to scan and potentially

T HE mobile WiMAX defined in 802.16e standards


addresses wireless broadband access for mobile users.
Key features added to fixed WiMAX standards include
handover to one of the neighbors, such as neighbor’s center
frequency, bandwidth, preamble id, etc. The MS also gets
handover related triggers through MOB_NBR-ADV message
support for Mobile Subscriber Station (MS) mobility through or another broadcast management message (DCD).
Handover [1-6]. Upon changes of radio link signal quality with serving BS
Handover support is a key feature that enables mobility in (e.g. RSSI or CINR change exceeds a preset value) that meets
WiMAX. Users and devices can roam across coverage area scan triggers, the MS initiates a neighbor scan. The MS
while maintaining not only their data connectivity but also negotiates with the serving BS scanning windows where it
active traffic. The standards define two types of handover: stops listening to the serving BS and then sync to and measure
hard handover and soft handover. Hard handover implements the neighbor BS signal quality. The MS and BS also negotiate
break-before-make, i.e., the Mobile Station (MS) breaks its the scan iteration, i.e., the number of scanning windows. This
radio link with the serving Base Station (BS) before exchange utilizes MOB_SCN-REQ and MOB_SCN-RSP
establishing radio link with the new target BS. In contrast, management messages. The scanning process results in traffic
soft handover is a make-before-break. The MS establishes a interruption, or delay on the downlink and uplink since the MS
radio link with the target BS before stopping its radio link with is consumed by scanning and measuring neighbor BS signal
the serving BS. Two methods of soft handover are defined: quality. The down time is controlled by the ability of the MS
Fast BS Switching (FBSS) and Macro Diversity Handover to sync to different neighbors and the total number of
(MDHO). In FBSS, service flows are set up with multiple neighbors, essentially the scan duration and the scan iteration.
BSs, only air link needs to be established for handover. In This can be expressed as
MDHO, service flows are established with a new anchor BS
before they are terminated with the previous anchor BS.
Tscan = NTs = αM
MDHO requires that two or more BSs transmit the same
service flows in the downlink (DL) and receive identical
service flows from the MS in the uplink (UL). Hard handover where N is total number of scan windows, Ts is the number of
is mandatory while soft handover is optional. In this paper, all frames in each scan window, M is the total number of
discussions and performance analysis are based on hard neighbors to be scanned and α is a coefficient related to the
handover. ability of the MS to sync to its neighbors.
Several studies have considered link performance for Handover can be initiated either by the MS or the serving
WiMAX and other networks with objective of determining BS. In case of MS initiated handover, the decision to start
impact of handover on traffic performance [7-11]. This paper handover is based on trigger conditions that are defined in the
is concerned with introducing a new methodology for studying DCD message sent by the BS. Upon meeting one or more of
handover performance and characterizing its behavior. these conditions, the MS starts the handover procedure by
sending MOB_MSHO-REQ to the serving BS. The serving
BS verifies that resources are available at the target BS
through backbone and prepares handover with the target BS.

978-0-7695-3719-1/09 $25.00 © 2009 IEEE 20


DOI 10.1109/MWS.2009.44
The serving BS confirms sufficient resources to meet Quality the arrival time is recorded and the difference is used to
of Service (QOS) parameters defined for the connections. The calculate data latency.
Serving BS sends MOB_BSHO-RSP to the MS. In case of BS Under normal operations the latency includes the latency of
initiated handover, the serving BS asks the MS to send scan the network including switches and routers as well as the
report. Based on the scan report and other conditions such as latency of the link between MS and BS. Since we are
traffic loading, the serving BS may direct the MS to handover interested in the latency of the WiMAX link and the minimal
to another BS by sending MOB_BSHO-REQ. possible latency is a WiMAX frame, we will measure all
Upon reception of either MOB_BSHO-RSP or latency results in units of WiMAX frame, i.e., dividing latency
MOB_BSHO-REQ, the MS notifies the BS with its intent to with frame length (5ms) and rounding the result.
start HO by sending MOB_HO-IND. Once the BS receives
MOB_HO_IND, or it believes the MS has left, it will stop
making DL allocation to the MS and stop polling the MS.
The MS syncs to the target BS and the latter provides UL
allocation to the MS to send RNG-REQ. Upon successful
reception of RNG-REQ, the target BS sends RNG-RSP to the
MS. Handover network re-entry is completed when MS
receives and processes the RNG-RSP message. (Note: other
optimized or non-optimized handover network re-entry are out
of scope and are not discussed in this paper.)

III. HANDOVER PERFORMANCE

Several performance metrics can be utilized to characterize


link performance in a wireless environment. These can be Figure 1: Experiment Setup
applied at the physical layer by measuring BER or application
layer through measurement of throughput, frame processing
rate and data latency. A. Scanning Performance
Due to the large number of MAC layer management To understand the latency performance during scan, the link
messages exchanged between the MS and BS and link quality to the serving BS is set such that the scan trigger is
interruptions (during scanning, ranging and network reentry), met. To prevent handover, the link to neighboring BSs is made
the most accepted performance metric to measure handover is worse than the link to the current serving BS. Thus, the
data latency. serving BS is the preferred station based on the handover
Data latency measures the total delay required for data metrics. So after one scan request, scan negotiation with BS
packets transmitted by the BS to be received by the MS. These and scan operation, the MS will stay with the original serving
measurements highlight normal operation of the system as BS. Since the scan trigger is still met, the MS will start
well as delays experienced by the packet due to unavailability another scan request with the serving BS. To prevent this from
of the MS. happening immediately, a scan holdoff timer is set. Only when
To fully characterize the system performance we propose an this timer expires, shall the MS start a new scan request.
experiment setup where all frames are occupied by data Figure 2 shows the data latency during the scan process.
packets in the uplink and downlink. The packet data size (P Graphs (a) and (b) in Figure 2 show the scanning frames as
bytes) is selected such that one packet is transmitted in a well as the traffic interruption during scan for DL and UL
single frame on DL and UL. This eliminates the impact of respectively. Here four scan requests and operations are shown
fragmentation and any scheduling interference both at MS and in the graphs. The scan holdoff period is 2000 frames, i.e., 10
BS. Since 200 frames are transmitted per second, the data rate second. Graph (c) in Figure 2 is a zoomed graph for the first
is equal to 200 × P × 8 × 10 −6 mbps. The packet protocol is scan operation of Graph (a) and (b). In this example, the
UDP. scanning window, i.e., the scan duration is equal to 5 frames
The setup is illustrated in Figure 1, where a traffic generator and there are 4 scan iterations in the scan operation. Graph (c)
that contains two interfaces (A & B) is used to exchange shows that 6 packets on the DL were transmitted after the link
packets between these interfaces. The packets traveling from resumes. This is because BS dumps all the packets buffered
A to B produce the UL traffic, i.e., traffic transmitted by MS during scan at the first frame when the link resumes. It also
to BS. Similarly, packets transmitted from B to A generate the shows that 11 packets have extra delay on UL compared to
DL traffic, i.e., traffic from BS to MS. At interface B, we have other packets. This is caused by MS being granted BW for UL
two BSs. They are neighbors of each other. Link quality to transmission every 4 frames in this case. After the scan, the
these two BSs is controlled by two attenuators so that MS may MS transmits packets held in the queue that contains 11
handover between them. packets. These 11 packets are transmitted in 2 frames, the first
For each packet generated, the transmitted time as well as contains 4 packets and the second contains 7 packets.

21
22

20

18

16

CINR
14

12

10
(a)
8
5 10 15 20 25 30 35 40
Time

Y BS1 BS 2

Figure 3: CINR changes for BS1 and BS2

With the CINR changes of BS1 and BS2, the MS handovers


(b)
between BS1 and BS2. Graph (a) of Figure 4 shows DL and
UL data latency during three such handover procedures.
Graphs (b) and (c) in Figure 4 show DL and UL data latency,
respectively, during one handover procedure in order to give a
close look at the traffic interruption during handover. The first
four traffic interruptions are due to scanning, which is
analyzed in the previous subsection. The last interruption is
due to the actual handover, i.e., the MS leaves the serving BS
and syncs to the target BS. This procedure starts after the MS
sends HO-IND to the serving BS and finishes after the MS
receives RNG-RSP from the target BS, and refreshes the
connection and security related information. The handover
procedure takes 16 frames in this example. Graph (b) shows
(c) that 17 packets on the DL were transmitted after handover is
completed. This is because BS dumps all the packets buffered
during handover at the first frame when the MS completes
Figure 2: Data latency during scan (a) DL, (b) UL, (c) DL and UL
handover to the target BS. On the UL side, Graph (c) shows
zoom
that after the MS completes handover to the target BS, the MS
receives two chunks of bandwidth, and it sends as many
packets as possible in the first allocation of bandwidth, then all
B. Handover Performance the remaining buffered packets are sent in the second
The handover is triggered by controlling link quality to the allocation of bandwidth. Due to the BS scheduling, BS cannot
serving BS and neighbor BSs. The attenuation on the link with allocate enough bandwidth for the MS to send all the packets
serving BS is increased to allow MS to scan neighbors, buffered during handover at one time. For both DL and UL, it
simultaneously, neighboring attenuation is reduced to ensure is shown that the largest data latency is around 20 frames, i.e.,
MS find a neighbor that meets handover trigger. Figure 3 100 ms, which in some applications might not be noticeable to
shows the CINR changes of BS1 and BS2. end user.

22
methodology populates every frame with data to capture link
behavior based on the latency experienced by data. DL and
UL data latency during scan and handover are captured and
analyzed. In a flawless wireless system, the data latency
caused by handover should be within expected limits and
unnoticeable to end user.
The advantage of the new method to measure handover
performance is apparent. Not only does it provide a measure
of the data latency during handover, it also provides insight
into the system performance and behavior.

(a)

REFERENCES

[1] IEEE P802.16e/D11: Amendment for Physical and Medium


Access Control Layers for Combined Fixed and Mobile Operation in
Licensed Bands, September 2005

[2] IEEE 802.16-2004: Air Interface for Fixed Broadband Wireless


Access Systems, October 2004

[3] WiMAX Forum, www.wimaxforum.org

[4] WiMAX Forum: Fixed, nomadic, portable and mobile


(b) applications for 802.16-2004 and 802.16e WiMAX networks,
November 2005

[5] WiMAX Forum: Mobile WiMAX – Part I: A Technical Overview


and Performance Evaluation, February 2006

[6] Eklund et, “WirelessMAN Insided the IEEE 802.16 Standard for
Wireless Metropolitan Networks”, IEEE Press, 2006.

[7] Choi S. “Fast handover scheme for real-time downlink services in


IEEE 802.16e BWA systems”. In: Proceeding of IEEE 61st
Vehicular Technology Conference. 2005. Vol.3: 2028~2032

[8] Kyung-ah K, Chong-Kwon K, Tongsok K, “A seamless handover


Mechanism for IEEE 802.16e Broadband Wireless Access”[C]. ISPC
2004, August 2004.

[9] Leung KK, Mukherjee S, Rittenhouse GE. “Mobility support for


IEEE 802.16d Wireless Networks”. WCNC 2005, 13-17 March,
2005: 1446~1452

[10] Lee D.H., Kyamakya K., Umondi J.P. “Fast handover algorithm
for IEEE 802.16e Broadband wireless access system”. ISWPC 2006
(c)
[11] Tae-Woon Kim; Sang-Hwa Chung; In-Su Yoon; Jeong-Soo Kim
Figure 4: Data latency during handover (a) DL and UL, (b) DL “Effects of Handover on TCP Congestion Control Algorithms over
zoom, (c) UL zoom Mobile WiMAX” 5th IEEE CCNC 2008. 5th IEEE, 10-12 Jan. 2008
Page(s):1230 - 1231
IV. Conclusion

In this paper, a new methodology using data latency is


introduced to characterize handover performance. The new

23

Das könnte Ihnen auch gefallen