Organization of the Dissertation

The rest of this dissertation is organized as follows. In Chapter 2, we survey the HO process in the IEEE 802.16e standard and review some existing IEEE 802.16e HO enhanced schemes. In

some existing MBS zone HO schemes. In Chapter 3, the proposed network architecture and the proposed NFHO scheme are described. In addition, we develop an analytic model for performance evaluation. In Chapter 4, we describe the proposed DMZ HO scheme for inter-MBS zone HO. An analytic model is also developed for performance evaluation. In Chapter 5, we give some concluding remarks and future work.

Chapter 2

Preliminaries and Related Work

In this Chapter, we survey the HO process in the IEEE 802.16e standard, and review existing IEEE 802.16e HO enhanced schemes. In addition, we brief the MBS data synchronization in the WiMAX standard [23] and review existing MBS zone HO schemes.

2.1 IEEE 802.16e HO Overview

2.1.1 Overview of the IEEE 802.16e HO Stages

The hard HO process consists of six stages: Cell Reselection, HO Decision and Initiation, Synchronization to Target BS Downlink, Ranging and Network Re-entry, Termination of MS Context, and HO Cancellation [2]. These stages can be functionally divided into two procedures:

HO Preparation and HO Execution. The HO Preparation procedure includes both Cell Reselection and HO Decision and Initiation stages. At the Cell Reselection stage, the MS requests the serving BS an allocation of scanning intervals. After the serving BS grants the scanning intervals, the MS maintains current connections with the serving BS, and then scan and synchronize with neighboring (NBR) BSs to evaluate the quality of each channel. An initial ranging procedure which is termed as association here is processed during a scanning interval with one of NBR BSs. The association procedure can obtain service availability information and ranging parameters, both of which will be used for HO target selection and expedite the HO

availability list of NBR BSs. At the HO Decision and Initiation stage, an HO decision is made for the MS to HO from the serving BS to the target BS.

The HO Execution procedure includes the following stages: Synchronization to Target BS Downlink, Ranging and Network Re-entry, Termination of MS Context, and HO Cancellation.

In this procedure, the MS starts actual HO. The MS synchronizes to the downlink (DL) of the target BS and obtains uplink (UL) parameters. After completing the ranging process, the MS obtains new basic and primary management CIDs (connection identifiers), which will be used for transporting management messages, and adjusts UL parameters to synchronize to the UL.

After synchronizing to the UL, the MS is capable of transmitting management messages. The MS then starts the Network Re-entry process that consists of Basic Capabilities Negotiation, Authorization, and Registration. Note that the ranging process in the Ranging and Network Re-entry stage is also termed as HO ranging. Moreover, in the registration of the Network Re-entry process, the target BS will reassign transport CIDs, which are used for transporting data packets, for active connections. After acquiring the transport CIDs, the MS is capable of starting data packet transmission. Termination of MS Context completes the HO process. In this stage, the serving BS releases the MS’s context after the Resource Retain Timer expires. In addition, the HO Cancellation stage is used to handle the situation of the MS canceling the HO before the Resource Retain Timer expires. In the HO Execution procedure, the data packet transmission is blocked until the MS acquires the transport CIDs in the registration of the Network Re-entry process. Fig. 2.1 shows two packet disrupted periods in IEEE 802.16e hard HO process. To shorten the second packet disrupted period so as to reduce the packet transmission delay in the HO Execution procedure, the transport CIDs assignment and UL synchronization issues must be resolved in advance. Therefore, the objective of Chapter 3 is proposing an HO enhanced scheme which can resolve the transport CIDs assignment and UL synchronization issues ahead of the Ranging and Network Re-entry stage.

NBR BS#2

MS Serving BS NBR BS #1

DL Synchronization

RNG-REQ / CDMA ranging code RNG-RSP(SBC-RSP, REG-RSP)

RNG_REQ / CDMA ranging code MOB_SCN-REQ

MOB_SCN-RSP

MOB_ASC-REP MOB_NBR-ADV

RNG_REQ / CDMA ranging code RNG_RSP information

RNG_RSP information

Packet disrupted

MOB_MSHO-REQ

MOB_BSHO-RSP

MOB_HO-IND

Packetdisrupted

Packets

Packets

CellReselection HO Decision& Initialization Sync. To TBS DL Ranging &Network reentry

Packets

Fig. 2.1: Packet disruptions in IEEE 802.16e hard HO Process [2]

During the HO Execution procedure, the serving BS buffers incoming DL packets. Upon completion of network re-entry, the target BS, which now is the new serving BS, will forward the data packets received from the old serving BS to the MS. The HO process will prolong the packet transmission delay because data packets are held during the HO Preparation and Execution procedures. Shortening the service disruption time caused by the HO process will minimize the impact on QoS. In Chapter 3, we focus on the IEEE 802.16e hard HO mode and propose a novel network architecture for IEEE 802.16e networks. Based on the network architecture, we design a network assisted fast handover (NFHO) scheme to accelerate the HO Execution procedure and to reduce the service disruption time resulted from the Ranging and Network Re-entry stage. As mentioned above, the DL packets forwarding from the serving BS to the target BS will prolong the packet transmission delay, especially in frequent HO situations.

packets to both the serving BS and target BS. It can also handle the problem of frequent HO of ping-pong mobility between BSs. An analytic model is developed to investigate the expected number of buffered packets, packet loss probability, and service disruption time during the HO Execution procedure. The analytic model is also used to analyze the performance among existing IEEE 802.16e HO enhanced schemes. We will show that the proposed NFHO scheme outperforms the existing IEEE 802.16e HO enhanced schemes. In addition, we will show that the packet loss probability is affected by the following network parameters: the HO service disruption time, the packet arrival rate of concurrent HO MSs, and the size of HO packet buffer pool. By the analytic model, we can evaluate the HO packet buffer pool utilization in a BS under different network parameter combinations and obtain a proper network parameter setting for meeting the QoS requirement. The analytic model can also be integrated to an admission control policy to provide proper QoS for incoming HO MSs.

2.1.2 IEEE 802.16e HO Scheme

In the IEEE 802.16e network, the BS periodically broadcasts network topology information via the MOB_NBR-ADV message. This message contains channel information of NBR BSs. Note that those message terms with all capital letters were defined in the IEEE 802.16e [2]. When an MS would like to HO, it starts the HO Preparation procedure and uses the MOB_SCN-REQ message to request a group of time intervals from the serving BS. Within the time intervals, the MS could seek and monitor a suitable BS from the list of candidate NBR BSs as the HO target.

Following the MOB_SCN-RSP message, the MS starts scanning and attempts to synchronize with each NBR BS to evaluate the quality of the channel. During the scanning intervals, all incoming/outgoing packets to/from the MS shall be buffered until exiting the scanning mode and returning to the normal operation mode. The duration of a scanning interval depends on which level (2, 1 or 0) of the association procedure is chosen. With Network Assisted

Association Reporting (association level 2) [2], the MS will not wait for an RNG-RSP message from each NBR BS after sending a RNG-REQ message or CDMA ranging code (for OFDMA) to the NBR BS. Instead, each NBR BS will send the serving BS the RNG-RSP message over the backbone. All RNG-RSP messages from each NBR BS are finally collected in an MOB_ASC-REP message, which will be sent to the MS by the serving BS [2]. Therefore, association level 2 needs a shorter scanning interval than the other levels. When the MS decides to HO, an MOB_MSHO-REQ message will be sent to the serving BS. After negotiating with the selected target BS, the serving BS sends the MS an MOB_BSHO-RSP message that may include an Action Time parameter to specify when the target BS will allocate a Fast Ranging IE (Information Element) [2]. The MS could use the Fast Ranging IE to transmit the RNG-REQ message, which expedites the HO ranging. The MOB_HO-IND sent by the MS is used to commit the HO. After committing the HO to the serving BS, the MS enters the HO Execution procedure. The MS proceeds to synchronize with the DL and then performs HO ranging, UL parameters adjustment, basic capabilities negotiation, authorization, and registration with the target BS. During the HO Execution procedure, the serving BS holds data addressed to the MS, and the target BS may use RNG-RSP to notify the MS of DL data pending. Once the MS registers to the target BS successfully, the target BS starts transmitting the retained DL pending data, forwarded from the serving BS, to the MS. After the MS re-establishes IP connectivity and completes reception of DL pending data, the target BS then use a backbone message to request the old serving BS to stop forwarding DL data. Note that our proposed NFHO scheme can restart DL/UL data transmission before the HO ranging, which can greatly reduce the service disruption time. In section 3.1, we will detail the proposed NFHO scheme.

2.2 Related Work for Enhancing IEEE 802.16e HO

The HO of wireless networks can be classified into vertical HO and horizontal HO. An HO is defined as vertical if it occurs between heterogeneous wireless networks. The work in [3][4]

proposes vertical HO decision algorithms to support HO between heterogeneous wireless networks. In contrast to vertical HO, an HO is defined as horizontal if it occurs between two adjacent cells of the same wireless network. In Chapter 3, we only focus on horizontal HO.

Existing IEEE 802.16e horizontal HO enhanced schemes are reviewed in this section.

2.2.1 Existing IEEE 802.16e HO Enhanced Schemes

During the HO process, real-time services may be disrupted. As shown in Fig. 2.1, in the HO Preparation procedure, the MS stops normal data transmission for the scanning of NBR BSs, which results in the first packet disrupted period. And in the HO Execution procedure, the normal data transmission is blocked until the MS completes the Ranging and Network Re-entry stage, which results in the second packet disrupted period. The blocking of data transmission disrupts the service of real-time traffic and increases packet transmission delay that impacts the QoS provision.

Existing IEEE 802.16e HO enhanced schemes focus on either layer-3 or layer-2. Layer-3 HO schemes, such as [5]-[8], are basically based on the IEEE 802.16e layer-2 hard HO scheme to accelerate layer-3 HO for mobile IPv6, and they did not reduce the packet transmission delay on layer-2. Therefore, these layer-3 HO schemes also have at least the same service disruption time as the IEEE 802.16e hard HO scheme. Existing layer-2 HO enhanced schemes can be functionally classified into improvements on either the HO Preparation procedure or the HO Execution procedure. The work in [9]-[13] focused on the HO Preparation procedure, and their algorithms are to predict an HO target BS to reduce ping-pong effects and the number of NBR BS scanning. In addition to HO target BS prediction, [9] also proposed a fast synchronization

and association scheme that makes the MS be able to do data transmission with the serving BS and to do association with the target BS simultaneously. To realize this scheme, it would costly base on either MDHO or FBSS mode. In addition, wrong target BS estimation, due to the exhaustion of target BS resources and so on, may cause huge delay when the MS repeats to evaluate a next target BS candidate.

The studies in [14] and [15] focused on reducing data latency in the HO Execution procedure. In [14], the authors proposed a Fast_DL_MAP_IE message to restart DL data transmission before the MS proceeds to the Ranging and Network Re-entry stage. The target BS will use Fast_DL_MAP_IEs and old CIDs for transmitting DL data packets immediately after the MS completes the Synchronization to Target BS Downlink stage. After HO ranging is competed and the REG-RSP message, including new transport CIDs assignment for active connections, is received, the MS will return to use normal DL_MAP_IE with new assigned transport CIDs. Similar to [14], the authors in [15] proposed another Transport CIDs assignment scheme to restart DL data transmission before the MS proceeds to the Ranging and Network Re-entry stage. Through its transport CIDs assignment scheme, the target BS can use the old transport CIDs, which were used in the serving BS, to transmit DL packets with no CID conflicts between the serving BS and the target BS. The MS uses old transport CIDs until it receives the REG-RSP message, which assigns new transport CIDs to active connections. Both schemes in [14] and [15] restart DL data transmission before the MS proceeds to the HO ranging; however, this feature is not applicable to the UL real-time traffic of these two schemes. This is because the scheme in [14] did not provide mechanisms to pre-acquire UL synchronization parameters, which is acquired during HO ranging, and new transport CIDs, which is acquired during registration. As to the scheme in [15], it can use old transport CIDs for UL data transmission until it receives the REG-RSP message; however, it did not provide mechanisms to pre-acquire

UL synchronization parameters. Thus, the scheme in [15] can advance UL data transmission only after HO ranging.

To restart UL data transmission, the MS should synchronize to the UL first. Since the RNG-RSP message provides UL synchronization parameters, such as frequency corrections, transmission power level corrections, UL timing offset corrections, and basic and primary management CIDs assignment, to advance UL data transmission, the UL parameters should be obtained ahead of schedule. In Chapter 3, we focus on reducing the service disruption time during the HO Execution procedure and propose an NFHO scheme that can restart both DL and UL data transmissions before the MS proceeds to the HO ranging. We will detail our scheme in section 3.1.

2.2.2 Qualitative Comparison of Existing IEEE 802.16e HO Enhanced Schemes

Focusing on the layer-2 IEEE 802.16e HO process, Table 2.1 shows a qualitative comparison of the existing IEEE 802.16e HO enhanced schemes. As mentioned above, existing layer-3 schemes were all based on the IEEE 802.16e layer-2 hard HO scheme. Therefore, from the layer-2 view, we regard the layer-3 schemes as the same as the IEEE 802.16e hard HO scheme.

Lee. et al. [9] is a representative of those schemes that focus on improving the HO Preparation procedure. Choi et al. [14] and Jiao et al. [15] focus on the HO Execution procedure, which is also the focus of the proposed NFHO scheme. Among existing IEEE 802.16e HO enhanced schemes, [15] is the only paper that dealt with the reduction of both DL and UL data latencies.

Our proposed NFHO scheme is also included in this qualitative comparison. Note that, to the best of our knowledge, our scheme is the only scheme that can restart data transmission at both DL and UL before the MS proceeds to the HO ranging.

Table 2.1: Comparison among existing IEEE 802.16e HO schemes.

HO Preparation procedure/

Reduce NBR BS association time

DL/UL after REG-RSP

Yes

Lee et al. [9]

HO Preparation procedure / 1. Reduce the number of NBR BSs scanning

2. Reduce association time

DL/UL after REG-RSP

Yes

Choi et al. [14]

HO Execution procedure / Reduce DL data latency

1. DL before

HO Execution procedure / Momentarily reuse old CIDs to reduce DL and UL data

HO Execution procedure / 1. Fast UL synchronization 2. Reduce DL and UL data

2.3 MBS in WiMAX Networks

2.3.1 WiMAX Standard for MBS

To concurrently transport common data to a group of mobile stations (MSs), the IEEE 802.16-2009 introduces multicast and broadcast service (MBS) in the downlink and provides macro-diversity and frame-level coordination modes among base stations (BSs) within an MBS zone. An MBS zone consists of a cluster of base stations (BSs) which transmit common MBS content with the same multicast connection identifier (MCID) and the same security association (SA). The backhaul MBS scheduler should coordinate all BSs within an MBS zone to synchronize MBS transmissions over radio interfaces. In an MBS zone, the set of medium access control (MAC) protocol data units (PDUs) carrying MBS content shall be identical in the same frame in all BSs [22]. To support intra-MBS zone data synchronization over radio interfaces, there are two options, macro-diversity and frame-level coordination, specified in the IEEE 802.16-2009 [22][23]. A macro-diversity MBS zone provides symbol level synchronization where the same MBS bursts are transmitted across involved BSs with time and frequency synchronized. An MBS zone, supporting frame-level coordination, provides frame-level synchronization, where the same MBS bursts are transmitted in the same frames across all involved BSs. In addition, to achieve MBS zone data synchronization, the WiMAX standard [23] defines a coordination mechanism to coordinate data transmission over the WiMAX network. The coordination mechanism includes a sync rule delivery procedure and recovery procedures. The sync rule delivery procedure announces the transmission timings of MBS bursts, and the recovery procedures consists of the sync rule recovery and the data path recovery, both of which are used to recover the lost sync rules and MBS payloads, respectively.

An MBS payload, including an MBS datagram, carries MBS content. The recovery procedures are based on the timeout and retransmission mechanisms.

In Chapter 4, we propose an in-frame control (IFC) scheme that aggregates MBS payloads and their associated sync rules. By this scheme, we can reduce the probability of entering recovery procedures so as to reduce packet transmission latency and packet buffer requirement.

In addition, to maintain maximum service continuity during inter-MBS zone HO, the WiMAX standard [23] defines a frame-offset coordination requirement for inter-MBS zone data synchronization. The frame-offset coordination defines that the same MBS service flow transmitted in any two adjacent MBS zones should be restricted to a specified number of frame-offset boundary [23]. The specified number of frame-offset boundary should be restricted within 7 frames [22][23]. Two levels of coordination requirements were defined. The level-1 coordination and level-2 coordination require service continuity and data continuity, respectively, between any two adjacent MBS zones. The service continuity focuses on providing non-interruption MBS service regardless of data discontinuity. The data continuity is a more strict form of service continuity; it provides not only service continuity but also non-interrupted MBS data reception. Consequently, from a data content perspective, the level-2 frame-offset coordination is the same with the frame-level coordination [23] , which provides HO MSs with

In addition, to maintain maximum service continuity during inter-MBS zone HO, the WiMAX standard [23] defines a frame-offset coordination requirement for inter-MBS zone data synchronization. The frame-offset coordination defines that the same MBS service flow transmitted in any two adjacent MBS zones should be restricted to a specified number of frame-offset boundary [23]. The specified number of frame-offset boundary should be restricted within 7 frames [22][23]. Two levels of coordination requirements were defined. The level-1 coordination and level-2 coordination require service continuity and data continuity, respectively, between any two adjacent MBS zones. The service continuity focuses on providing non-interruption MBS service regardless of data discontinuity. The data continuity is a more strict form of service continuity; it provides not only service continuity but also non-interrupted MBS data reception. Consequently, from a data content perspective, the level-2 frame-offset coordination is the same with the frame-level coordination [23] , which provides HO MSs with