Broadband Video Streaming during Backhaul WiMAX Handovers Salah S. Al-Majeed and Martin Fleury, Member, IEEE

Abstract — Broadband Video Streaming (BVS) with selective retransmission trades a reduced but acceptable video quality during IEEE 802.16e (WiMAX) hard handover (HHO) for improved end-toend latencies. BVS comes in two forms: selective and unselective retransmission when packet loss occurs. Both forms promise better video quality if an HHO occurs than UDP transport or traditional congestion-controlled streaming. If a vertical handover occurs, the unselective form of BVS is preferable. Both variations are intended for IPTV content management networks, which reduce BVS retransmission latency by the proximity of the server1. Index Terms — WiMAX, video streaming, hard handover, vertical handover

I. INTRODUCTION An IEEE 802.16e (mobile WiMAX) [1] can act as a backhaul network for an on-bus IEEE 802.11 network. As trialled in Stockholm, Sweden, when a bus travels along its fixed route, passengers access the Internet via a set of WiMAX masts, with horizontal handover2 occurring as the bus moves between masts. As a possible extension, when a passenger with a multi-homed device alights from the bus and walks away, a vertical handover (VHO) takes place between an on-bus 802.11 access-point (AP) and another WiMAX base station (BS) in the vicinity. VHO can be accomplished with the Media Independent Handover (MIH) of the IEEE 802.21 standard [2]. Mobile Internet Protocol TV (IPTV) [3] is an attractive additional service for urban commuters but there is a possibility of interruption during these transitions due to: route setup delay; signalling message overhead and processing time; and packet loss. To guard against packet loss, this paper proposes a simple negative-acknowledgment (NACK) scheme which for naming convenience is called Broadband Video Streaming (BVS). In the variety examined (BVS-I), for horizontal handover, only intra-coded I-frame packets when lost are retransmitted. This restriction has the effect of reducing streaming interruption during a hard handover (HHO). In contrast, during a VHO, experiments indicate that BVS without selection is preferable, leading to a recommendation that a hybrid scheme is employed to cope with the possibility of either types of handover occurring in the course of a streaming session. The proposed transport scheme, whether BVS or BVS-I, assumes a remote video server and end-to-end control, without the need for cross-layer intervention at the access network. It

could be applied directly to WiMAX-to-vehicle communication and, in principle, to any broadband wireless technology with a fast feedback channel, including IEEE 802.16m with network-directed HHO [4]. During VHO, even though IEEE 802.21 attempts to harmonize signaling, the extent of signally differs between the two network types, because of the relative complexity of WiMAX, which adds quality-of-service management to the underlying transmission system. The WiMAX forum specifies just 50 ms horizontal handover delay. However, simulations in this paper suggest that, because of the presence of channel errors, the overall effect (not just the immediate transfer) [5] may be unsatisfactory for UDP-transported video streams [6] or transport through standardized TCP-friendly Rate Control (TFRC) [7]. The simulations were conducted with the National Institute of Standards and Technology (NIST) mobility add-on for the well-known ns-23 simulator, together with the NIST add-on for IEEE 802.16, which includes IEEE 802.16e scanning and handover support. Given the deficiencies of some existing transport schemes, a further contribution of the paper is a transport method tuned to the needs of IPTV, which paves the way for new services, such as time-shifted TV and video-on-demand, which can follow the user. The remainder of this paper is organized as follows. Section II briefly reviews ways to improve latency during handover and the contribution of IEEE 802.21 to harmonizing VHO across wireless technologies. Section III considers BVS and its selective variety BVS-I alongside a traditional congestion control method, namely TFRC. We have conducted detailed simulations and Section IV describes the settings for these. Section V presents our findings for horizontal and vertical handover, while Section VI concludes the paper by reflecting on the findings. II. HANDOVER MECHANISMS While horizontal handover is concerned with migration between homogeneous networks, VHO is more intricate as it involves signaling between heterogeneous networks. Handovers are either soft, in which the previous connection is kept alive until the new connection is made, or hard, in which the previous connection is broken before the current one is made. Handovers can be: entirely controlled by a mobile device; be assisted by the mobile device though executed by the access network, based on connection information at the mobile; or initiated by the access network, without any action

1

S. Al-Majeed and M. Fleury are with the University of Essex, Colchester, CO4 3SQ UK (e-mail: {ssaleha, [email protected]). 2 We use the terms “handover” and “handoff” interchangeably in this paper.

3 Available from http://w3.antd.nist.gov/seamlessandsecure/ [accessed July 2010]

by the mobile device. Handover consists of: detection of a new access network and selection of that network; resource allocation as a new connection is established; and the update of routes and forwarding of data over the new connection. Scanning is the process of acquiring information about neighboring access networks by the mobile device. In IEEE 802.16e, exchange of information for this purpose takes place between Base Stations (BS’s) over the backbone network. IEEE 802.16j introduced relay stations [8], which are simplified BS’s with radio connection only but the simulation is restricted handovers between BS’s. Mobile WiMAX supports three handover mechanisms but only the mandatory HHO at layer 2 can be accomplished with a single channel at any one time, thus reducing equipment cost and improving BS capacity. HHO employs a break-beforemake procedure which reduces signalling. As is normal, a mobile station (MS) monitors signal strength from adjacent BSs, employing a hysteresis mechanism to avoid thrashing between BS. The MS must then: obtain uplink and downlink parameters; negotiate capabilities; gain security authorisation and exchange keys; register with the BS; and establish connections. It is expected that these mechanisms, whether for horizontal or vertical handover, will be subsumed in the emerging IEEE 802.11.21 [2] standard. IEEE 802.21 specifies tools to exchange information, commands, and events but does not standardize the execution mechanism. The architecture of IEEE 802.2’s MIH appears in Fig. 1. In this paper for mobility management, mobile IP (MIP) is assumed rather than the Session Initiation Protocol (SIP). Mobile IP acts as an upper layer client of 802.21’s MIH function (MIHF). The MIHF itself lies between layer 2 (Datalink — Medium Access Control (MAC)) and layer 3. Layers 3 and above can obtain information, receive event notifications, and issue commands via MIH, while the MIHF provides a Service Access Point to layer 2 and below. Network information includes MAC addresses, security information, and channel information. Events include link parameter changes and link status changes. In the successor to IEEE 802.11e, IEEE 802.16m [2], handover is hard and network controlled. The role of the MS is confined to suggesting alternative BS’s if a connection to the target BS fails. Reduced authentication procedures may occur after negotiation between the BS and the target BS. Consequently, in seamless handover, a MS may exchange data packets with the target BS before a network re-entry control transaction. In addition, entry-before-break allows connection with the original BS while at the same time negotiating with the target BS. At no point are there two data paths at the MS. There are several ways to improve handover management for real-time services. The first way is to make structural changes to the way a handover operates such as reducing the latency of the network selection process [9] and/or the mobility management [10]. It is also possible to act at the application layer through increased protection against packet loss and delay. If the handover can be anticipated then prebuffering [11] at the client is possible. In [12], it is noted that,

SIP

MIPv4

text

MIPv6

…..

Higher layer (3 and above) MIH events

MIH commands

Information service

MIH Function text Link events 802.3

Link commands

Information service

802.11

3GPP

802.16 text

3GPP2

Lower layer (2 and below)

Fig. 1. Architecture of IEEE 802.21’s MIH.

receiver notification of increased packet losses and round-trip times are insufficient handover indicators, because they occur after the event. Instead, in [12] information about an impending handover is passed up the protocol layers. Alternatively, this paper seeks to adapt the transport scheme to the needs of handover and video streaming. The advantage of this second way is that it neither alters the way handovers are controlled nor requires special intervention for video applications. III. IPTV TRANSPORT Various methods of improving upon UDP offer the possibility of improved media transport without the overhead of application layer congestion control superimposed upon UDP transport. This is particularly the case if the latency between the streaming server and the mobile device is relatively small. Because content management can bring the server closer to the access network [3], reduced latencies are likely to occur. In fact, one industry source recommends [13] a maximum cumulative delay factor of between 9 and 50 ms for IPTV delivery over a network. Unembellished UDP has been used for IPTV transport over IEEE 802.16e systems [6]. However, UDP packet losses can seriously harm a compressed video stream. Bell Labs. introduced a reliable form of UDP, R-UDP, see [14], and there is also a coincidently-named R-UDP protocol employed by Microsoft for IPTV service delivery over multicast networks. A. Operation of TFRC A common alternative to direct use of UDP is a standardsbased form of transport, which supplements the underlying UDP transport by application layer congestion control. TFRC [7] or Datagram Congestion Control Protocol (DCCP) [15], which adds connection set-up to TFRC, have been used directly [16] or in cross-layer form [17] for video streaming over wireless networks. As described in [7], TFRC is a receiver-based system in which the packet loss rate is found at the receiver and fed-back to the sender in acknowledgment messages (for reliability through TCP in Section IV’s simulations). The sender calculates the round-trip time from the acknowledgment messages and updates the packet sending rate. A throughput equation models TCP New Reno to find the sending rate:

TFRC (t rtt , t rto , s, p )  t rtt

s  2bp 3bp   p (1  32 p 2 )  t rto min1,3 3 8  

(1) where trtt is the round-trip time, trto is TCP’s retransmission timeout, s is the segment size (TCP’s unit of output) (herein set to the packet size), p is the normalized packet loss rate, wm is the maximum window size, and b is the number of packets acknowledged by each ACK. b is normally set to one and trto = 4trtt. It is important to notice that trto comes to dominate TFRC’s behavior in high packet loss regimes. Clearly packet loss and round-trip time cause the throughput to decrease in (1), whereas other terms in the denominator are dependent on these two variables.

Video Application H.264 Video Encoder

H.264 Video Decoder

H.264 Video Coding Layer (VCL) Pic. I

Pic. P

Pic. B

Network Abstraction Layer (NAL) Re-order RTP packet

RTP packet

Buffer

Buffer

No NACK Ret. lost packet?

Yes

No

Packet Type = A?

Yes No Packets Queue

Yes

Initial packet loss?

Scheduler

B. Operation of BVS For BVS-I, in Fig. 2, at an MS, a record is kept of packet sequence numbers available through the Real Time protocol (RTP) header and, if an out-of-sequence packet arrives from the server, a NACK may be transmitted to the BS in the next 802.16e sub-frame to be forwarded to the video server. The MS only transmits a NACK if this is the first time that particular packet has been lost. To reduce the overhead at the MS, the decision as to whether to retransmit a packet is left to the server. The server prevents transmission from its input buffer until a single retransmission of the missing packet in the sequence has taken place. Not shown in Fig. 2, is a holding buffer that retains sent packets in anticipation of the need for a retransmission. However, the server will only retransmit if the NACK refers to an I-frame packet. In Fig. 2, the packet is defined as of type A, as other forms of prioritized retransmit are possible, and, in fact, all packets are retransmitted in simple BVS. IV. EVALUATION PROCEDURE In this Section, we detail the WiMAX and WiFi (IEEE 802.11) simulation configurations, along with other information about the simulations such as the wireless channel model and the video characteristics. A. Simulation configuration Simulations were conducted for IEEE 802.16e (mobile WiMAX) with horizontal handover to another BS and with vertical handover from and to an IEEE 802.11 WLAN. IEEE 802.21 was modelled with the NIST handover add-on for ns-2, which is tied to the IEEE 802.11b model built into ns-2 (hence the use of this version of IEEE 802.11 in simulations). 802.11b operated at 11 Mbps. 25 runs per data point were averaged (arithmetic mean) and the simulator was first allowed to reach steady state before commencing testing. In Fig. 3’s horizontal handover scenario, a remote server at C streams video over the IP network, while node A sources to node B constant bitrate (CBR) data at 1.5 Mbps with packet size 1 kB. Node A also sinks a continuous TCP FTP flow sourced at node B. Node B sources an FTP flow to the BS and CBR data at 1.5 Mbps with packet size 1 kB. The MS moves in parallel to the two BS, which are separated by 1.9 km. For

Transmission

Receiver

Wire / Wireless IP Network

Fig. 2. Operation of BVS.

VHO, the same scenario was employed but an IEEE 802.11 AP substituted for BS2 in Fig. 3. B. Wireless configuration To evaluate the proposal, transmission over WiMAX was carefully modeled. The PHYsical layer settings selected for WiMAX simulation are given in Table I. The antenna heights and transmit power levels are typical ones taken from the Standard [1]. The antenna is modeled for comparison purposes as a half-wavelength dipole, whereas a sectored set of antenna on a mast might be used in practice to achieve directivity and, hence, better performance. Similarly, Multiple-Inpute Multiple-Output (MIMO) antennas are not modeled. The IEEE 802.16 Time Division Duplex (TDD) frame length was set to 5 ms, as only this value is supported [18] in the WiMAX forum adaptation of the Standard. The data rate results from the use of one of the mandatory coding modes [1] for a TDD downlink/uplink sub-frame ratio of 3:1. The BS was assigned more bandwidth capacity than the uplink to allow the WiMAX BS to respond if necessary to multiple mobile devices. Thus, the parameter settings in Table I such as the modulation type and PHY coding rate are required to achieve a datarate of 10.67 Mbps over the downlink. Buffer sizes were set to 50 packets (a single MAC Service Data Unit within a MAC Protocol Data Unit). This buffer size was selected as appropriate to mobile, real-time applications for which larger buffer sizes might lead both to increased delay and larger memory energy consumption in mobile devices. As a point of comparison, capacity studies [18] suggest up to 16 mobile TV users per mobile WiMAX cell in a ‘lossy’ channel depending on factors such as the form of scheduling and whether MIMO is activated. Settings for the IEEE 802.11 AP are given in Table II, whereas the operation of 802.11 is assumed to be well-known.

C

BS1 MS

DL

IP Core Network

DL Handoff C BS2 100 Mbps 2 ms

5 Mbps 2 ms

R

R

100 Mbps 2 ms 100 Mbps 2 ms

100 Mbps 2 ms A

B

Fig. 3. Video streaming handover scenario. TABLE I.

IEEE 802.16E PARAMETER SETTINGS

Parameter PHY Frequency band Bandwidth capacity Duplexing mode Frame length Max. packet length Raw data rate (downlink) IFFT size Modulation Guard band ratio MS transmit power BS transmit power Approx. range to MS Antenna type Antenna gains MS antenna height BS antenna height Receiving threshold

Value OFDMA 5 GHz 10 MHz TDD 5 ms 1024 B 10.67 Mbps 1024 16-QAM 1/2 1/16 245 mW 20 W 1 km Omni-directional 0 dBD 1.2 m 30 m 7.91e-15 W

OFDMA = Orthogonal Frequency Division Multiple Access, QAM = Quadrature Amplitude Modulation, TDD = Time Division Duplex TABLE II.

IEEE 802.11B PARAMETER SETTINGS

Parameter PHY Frequency band Bandwidth capacity Max. packet length used Raw data rate (downlink) AP transmit power Approx. range Receiving threshold

Value DSSS 2.4 GHz 20 MHz 1024 B 11 Mbps 0.0025 W 100 m 6.12e-9 W

DSSS=Direct-Sequence Spread Spectrum

C. Wireless channel model ‘Bursty’ errors as occur during fast fading were generated by the Gilbert-Elliott model [19], which is a form of hidden Markov model with internal good and bad states. If PGB and PBG are the probabilities of going from the good to bad state and from going from the bad to good state respectively, then

πG = PBG/(PBG + PGB)

(2)

πB = PGB/(PBG + PGB)

(3)

are the steady state probabilities of being in the good and bad states. The probability of remaining in the good state, PGG, was set to 0.95 and of remaining in the bad state, PBB, was 0.94, with both states modeled by a Uniform distribution. The packet loss probability in the good state, PG, was fixed at 0.01 and the bad state probability, PB, was made variable. The Gilbert-Elliott model was selected, as it is the presence of burst errors [20] that mostly affects the quality of compressed video. D. Video configuration As a test, the Paris sequence H.264/AVC (Advanced Video Coding) codec [21] Variable Bit-Rate (VBR)-encoded at 30 frame/s with Common Intermediate Format (CIF) (352  288 pixel/frame) with quantization parameter (QP) set to 26 (from a range 0 to 51). The Peak-Signal-to-Noise Ratio (PSNR) for this sequence without packet loss is 38 dB. The slice size was fixed at the encoder at 900 B. In this way the risk of network segmentation of the packet was avoided. Consequently, the risk of loss of synchronization between encoder and decoder was reduced. Paris consists of two figures seated round a table in a TV studio setting, with high spatial-coding complexity and moderate motion. Quality-ofexperience tests show [22] that this type of content is favored by users of mobile devices as it does not stretch the capabilities of the screen display. The Intra-refresh rate was every 15 pictures with an IPBB…I coding structure. 1065 frames were transmitted resulting in a video duration of 35.5 s. Simple previous frame replacement was set for error concealment at the decoder as a point of comparison with others’ work. Other forms of error concealment increase decoder complexity. V. FINDINGS In this Section, we compare the performance of the various transport schemes during horizontal and vertical handover. A. Performance during horizontal handover In Figs. 4 and 5, BVS and BVS-I’s objective video quality at the MS are compared for increasing speeds and variable channel conditions. At around 20 mps (45 mph), BVS-I’s quality becomes unsatisfactory as it drops below 25 dB. At low bus speeds and shorter error bursts both BVS and BVS-I deliver ‘good’ quality video (above 31 dB). From Fig. 6, both BVS and BVS-I are superior to UDP-transport and TFRC. From the summary in Table III, without retransmissions, packet loss approaches 10%. TFRC reduces its sending rate by increasing the inter-packet gap but only by approximately doubling the sending period of Paris. BVS is better but its latencies are larger. However, by reduced retransmission with BVS-I, end- to-end latencies are reduced. Simulation has

TABLE III.

SUMMARY COMPARISON AT 10 MPS (22 MPH) WITH PB = 0.15

45 10 mps 15 mps 20 mps

Throughput ( kbps) Sending period ( s ) Packet Loss ( % ) Jitter ( s ) Mean end-to-end delay ( s ) Max. end-to-end delay (s ) PSNR ( dB ) S.D. (dB)

PSNR ( dB )

40

35

30

UDP 762.9 35.43 9.75 0.008 0.008 0.256 25.29 3.28

TFRC 371.1 72.97 7.29 0.066 0.007 0.247 27.26 3.31

BVS 819.2 35.70 1.69 0.007 0.015 0.631 34.82 3.42

BVS-I 766.4 37.49 3.81 0.008 0.011 0.329 29.05 4.91

25

20 0

0.05

0.1

0.15

0.2

0.25

Wireless Channel Error Rate ( PB )

Fig. 4 Y-PSNR of BVS for different channel conditions. 45 10 mps 15 mps 20 mps

PSNR ( dB )

40

35

30

25

20 0

0.05

0.1

0.15

0.2

0.25

Wireless Channel Error Rate ( PB )

Fig. 5. Y-PSNR of BVS-I for different channel conditions. 45 UDP TFRC BVS BVS-I

PSNR ( dB )

40

35

30

25

20 0

0.05

0.1

0.15

0.2

0.25

Wireless Channel Error Rate ( PB )

Fig. 6. Video quality from different protocols with a speed of 10 mps (22 mph).

shown that at speeds above 45 mps (100 mph), HHO latency grows rapidly. B. Performance during vertical handover Frame-by-frame results in Fig. 7 isolate the effect on video quality (PSNR) of vertical handovers from IEEE 802.16e to 802.11b and vice versa. When the MS loses connection to the IEEE 802.16e BS it is at 0.93 km distance from the BS, whereas when it connects to the IEEE 802.11b AP it is at 70 m distance from the AP. For ease of analysis, no other mobile

station is present in either the IEEE 802.16e or the 802.11b network. By default, the speed of movement of the MS was 1 m/s (2.2 mph), i.e. as appropriate for somebody strolling from the vicinity of the BS to the vicinity of the AP while using a mobile device. Notice that, in a short study of the impact of different speeds [23] when going between these networks, it was found that IEEE 802.18e’s performance actually improved at higher speeds (up to 10 mps) because of IEEE 802.16e’s superior mobility management. Plots shown in Fig. 7 are unembellished UDP, TFRC, and the two varieties of the lightweight NACK scheme: that is BVS when all initially lost packets are NACKed and BVS-I when only I-picture bearing packets are NACKed. Because of the longer exposure of packets to channel conditions, video quality is generally worse during an IEEE 802.16e connection than a connection to 802.11. BVS is clearly better able to cope with the transition in Fig. 6a, while BVS-I only results in a limited gain over UDP. TFRC begins to recover quality after the handover but it does not compete with BVS. Notice that during the stable period prior to a handover from 802.11 to 802.16e, re-sending only I-frames (BVS-I) is sufficient to maintain quality, with reduced throughput and reconstruction delay at the decoder. However, during worse conditions under 802.16e prior to a handover, Fig. 7b, BVS-I results in lower quality than BVS, as do the other two options. In an extended test, the MS journeyed from the BS to the AP and back again. An approximately equal time was spent under IEEE 802.16e and IEEE 802.11b streaming. As Table IV shows for the two handover scenario, the TFRC response to both congestion and channel errors is to increase the interpacket gap such that the total sending period of the 34.5 s clip grows to an unacceptable level (about 8 s longer than the display time), though all transport methods suffer from interruptions (freeze frame effects). TFRC also suffers from poor wireless channel utilization, as its throughput declines. Both the proposed schemes result in reduced packet loss, i.e. the packets lost even after a single attempt at retransmission. A consequence of retransmission is greater end-to-end packet delay, when delay is the aggregate of an initial transmission and any delay from resending a packet after a NACK. However, the delay in both cases is less than 50 ms. The main deficiency of BVS appears to be the delay that occurs during handovers, as the maximum end-to-end delay is high. However, it is actually the maximum period of interruption caused to frame display rather than the impact of delay from individual packets making up the frame that is

40

40 TFRC

35

PSNR ( dB )

PSNR ( dB )

UDP

30 25 20 0

20

40

25

Frame Index ( x 20 ) 40

SUMMARY AT SPEED 2 MPS OF DIFFERENT TRANSPORT SCHEMES AFTER TWO VERTICAL HANDOVERS — THE FIRST FROM IEEE 802.16E TO 802.11B AND THE SECOND FROM IEEE 802.11B TO IEEE 802.16E WITH DATA THE MEAN OF 25 SIMULATION RUNS, PB = 0.15, SPEED 2 M/S

30

20 0

60

TABLE IV.

35

20 40 Frame Index ( x 20 )

60

40

35

PSNR ( dB )

PSNR ( dB )

BVS-I

30 25 20 0

BVS 20 40 Frame Index ( x 20 )

35 30

UDP 838 35.48 3.84 0.0077

TFRC 687 43.08 3.29 0.0093

BVS 851 35.79 0.38 0.0075

BVS-I 855 35.58 2.49 0.0076

0.0141

0.0106

0.0150

0.0149

0.0680

0.0520

0.2980

0.0790

31.29 5.85 0.301

32.31 6.43 0.309

37.43 2.52 0.300

33.01 6.09 0.301

25 20 0

60

Throughput ( kbps) Sending period ( s ) Packet loss ( % ) Packet jitter ( s ) Mean packet end-to-end delay ( s ) Max. packet end-to-end delay (s ) PSNR ( dB ) Standard deviation (dB) Max. interruption (s)

20 40 Frame Index ( x 20 )

60

PSNR ( dB ) 40 UDP TFRC BVS BVS-I

40

40

35

35

PSNR ( dB )

PSNR ( dB )

(a)

30 25

35

30 30

25 TFRC

UDP 20 40 Frame Index ( x 20 )

20 0

60

40

40

35

35

PSNR ( dB )

PSNR ( dB )

20 0

30 25

20 40 Frame Index ( x 20 )

20 40 Frame Index ( x 20 )

25 0

1

2

3

4

Fig. 8. Variation of video quality for the different schemes with differing speed.

30

PSNR ( dB ) 25

40 UDP PSNR BVS BVS-I

BVS-I 60

5

Speed ( mps )

BVS 20 0

60

20 0

20 40 Frame Index ( x 20 )

60

(b)

35

Fig. 7. Frame-by-frame video quality (a) during vertical handover from IEEE 802.11b to 802.16e (b) from IEEE 802.16e to 802.11b.

significant. In that respect all systems appear to behave similarly. An important point to note is that packet loss during handover is heavily dependent on speed, even at slow walking speeds. The packet loss reflects itself in the overall video quality which is shown for different speeds in Fig. 8. Notice that 2 mps or 4.5 mph is already fasting than walking speed of about 3.3 mph. Likewise, see Fig. 9, variation in channel condition can notably affect video quality, especially during communication with the IEEE 802.16e BS, because of the longer transmission times involved. However, though BVS suffers from poor channel conditions its video quality overall remains superior to all the other schemes, including BVS-I.

30

25 0

0.05

0.1

0.15

0.2

0.25

0.3

Wireless Channel Error Rate ( PB )

Fig. 9. Variation of video quality for the schemes with differing channel quality.

VI. CONCLUSION Results show that at moderate speeds video quality during a WiMAX horizontal handover could improve by as much as 9 dB using BVS instead UDP but there is a cost in end-to-end delay. This issue is resolved in BVS-I by selective retransmission by picture type, while video quality remains acceptable. The findings do not suggest acceptable streaming from a remote server at speeds much over 20 mps (45 mph), which is probably the maximum bus speed in large cities. The lightweight transport method is sufficient in the presence of network congestion along the path from the video server to the mobile device. TFRC, which requires an acknowledgment after every packet transmission, is more affected by congestion in the feedback path than the BVS scheme. Just as for TCP, TFRC is also affected by its inability to distinguish between those packet losses due to congestion (on the streaming path) and those due to packet drops on the wireless channel. An interesting observation from the simulations is that BVS-I’s video quality was markedly inferior to BVS during a VHO. However, during BVS-I streaming to the IEEE 802.11 AP prior to handover good video quality resulted with less packet delay despite reduced throughput. This suggests that handover detection at an MS will make a hybrid BVS/BVS-I scheme effective. REFERENCES [1]

IEEE, 802.16e-2005, IEEE Standard for Local and Metropolitan Area Networks. Part 16: Air Interface for Fixed and Mobile Broadband Wireless Access Systems, 2005. [2] K. Taniuchi et al.: “IEEE 802.1: Media Independent Handover: Features, applicability, and realization,” IEEE Commun. Mag., vol. 47, no. 1, pp. 112-120, 2009. [3] N. Degrande, K. Laevens, and D. De Vleeschauwer, “Increasing the user perceived quality for IPTV services,” IEEE Commun. Mag., vol. 46. no. 2, pp. 94-100, 2008. [4] R.Y. Kim, I. Jung, X. Yang, and C.-C. Chou, “Advanced handover schemes in IMT-Advanced systems,” IEEE Commun. Mag., vol. 47, no. 8, pp. 68-85, 2010. [5] K. Daniel, S. Rodhe, S. Šubik, and C. Wierfeld, “Performance evaluation for mobile WiMAX with a continuous scanning algorithm,” in IEEE Mobile WiMAX Symp., 2009, pp. 30-35. [6] O. Issa, W. Li. and W. Liu, ‘Performance evaluation of TV over broadband wireless access networks,” IEEE Trans. Broadcast., vol. 56, no. 2, pp. 201-210, 2010. [7] M. Handley, J. Pahdye, S. Floyd, and J. Widmer, “TCP-Friendly Rate Control (TFRC): Protocol Specification,” IETF, RFC 3448, 2003. [8] Z. Becvar, P. Mach, and R. Bestak, “Optimization of Handover procedure in WiMAX networks with realy stations,” in IEEE Int. Symp. on Wireless Pervasive Computing, 2008, pp.581-584. [9] S.K. Lee, K. Sriram, K. Kim, Y.H. Kim, and N. Golmie, “Vertical handoff decision for providing optimized performance in heterogeneous networks,” IEEE Trans. Veh. Technol., vol. 58, no. 2, pp. 865-881, 2009. [10] K.-S. Kong, W. Lee, Y.-H. Han, and M.-K. Shin, “Handover latency analysis of a network-based localized mobility management protocol,” in IEEE Int. Conf. on Commun., 2008, pp. 5838-5843. [11] D. Lee, J.W. Kim, and P. Sinha, “Handoff-aware adaptive media streaming in mobile IP networks,” in Int. Conf. on Info. Networking, 2006. [12] S. Ishihara, K. Koyama, G. Miyamoto, and M Kuroda, “Predictive rate control for realtime video streaming with network triggered handover,” in IEEE Wireless Commun. and Networking Conf., 2005, pp. 13351340.

[13] Agilent Technologies, “Validating IPTV service quality under multiplay network conditions,” White Paper, 2008. [14] C. Partridge, and R. Hinden, “Version 2 of the Reliable Data Protocol,” Internet Engineering Task Force, RFC 1151, 1990. [15] E. Kohler, M. Handley, and S. Floyd, “Datagram Congestion Control Protocol,” Internet Engineering Task Force, RFC 4340 , 2006. [16] K. Tappayuthpijam, G. Liebl, T. Stockhammer, and E. Steinbach, “Adaptive video streaming over a mobile network with TCP-Friendly Rate Control,” in Int. Conf. on Wireless Commun. and Mobile Computing, 2009, pp. 1325-1329. [17] B. Görkemli, M.O. Sunay, A.M. Tekalp, “Video streaming over wireless DCCP,” in IEEE Int. Conf. on Image Processing, 2008, pp. 2028–2031. [18] C. So-In, R. Jain, and A.-K. Tamimi, “Capacity evaluation for IEEE 802.16e mobile WiMAX,” Journal of Computer Systems, Networks, and Commun. [online], 12 pages, 2010. [19] G. Haßlinger, and O. Hohlfeld, “The Gilbert-Elliott model for packet loss in real time services on the Internet,” in 14th GI/ITG Conf. on Measurement, Modelling, and Evaluation, 2008, pp. 269–283. [20] Y.J. Liang, J.G. Apostolopoulos, and B. Girod, “Analysis of packet loss for compressed video: Effect of burst losses and correlation between error frames,” IEEE Trans. Circuits Syst. Video Technol., vol. 18, no. 7, pp. 861-874, 2008. [21] T. Wiegand, G.J. Sullivan, G. Bjontegaard, and A. Luthra, “Overview of the H.264/AVC video coding standard,” IEEE Trans. Circuits Syst. Video Technol, vol. 13, no. 7, pp. 560-576, 2003. [22] F. Agboma, and A. Liotta, “Addressing user expectations in mobile content delivery,” Mobile Information Systems, vol. 3, nos. 3/4, pp. 153-164, 2007 [23] M. Thaalbi, and N. Tabbane, “Vertical handover between WiFi network and WiMAX network according to IEEE 802.21 standard,” In K. Elleithy et al. (eds.) Technological Developments in Networking, pp. 533-537, 2010.