MEASUREMENTS FOR HANDOFF OPTIMIZATION

Handoff optimization is a key challenge for the network management as it results in the enhancement of the network performances by optimizing throughput, routing, delay profiles, delivered QoS, and communication costs. Therefore, mobile WiMAX should take into consideration numerous parameters that implement a particular handoff policy to optimize the handoff performance, especially in case of interaction with heterogeneous networking technologies. It is valuable to note that IEEE 802.16e specifications consider the handover decision algorithms beyond the scope of the standard.

1 Monitoring Parameters

Radio resource management (RRM) specifications should guarantee efficient resource utilization in WiMAX networks by assisting functions such as QoS provision, service flow admission control, and mobility management. RRM mechanisms shall be implemented in ASNs either with BSs that directly communicate between them or with BSs having no direct communication between them or at a centralized RRM entity that does not reside in a BS but that collects and monitors radio resource indicators from several BSs. Each BS should collect data about the neighboring BSs either through static configuration data or through a different RRM entity aware of the dynamic load of neighboring BSs. RRM mechanisms should support network functions in taking the required decisions. For instance, RRM may assist handoff preparation and control to improve the overall performances. 

For example, RRM mechanisms may optimize the system load control by selecting the most suitable target BS during handoff or updating the list of recommended BSs to handoff. The radio resources available at a BS where a BS-ID defines a sector with a single frequency assignment may be a good hint for BS selection during network entry or handoff. More specifically, the available radio resource indicator, which gives the percentage of reported average available subchannels and symbols resources perframe, may be used to select recommended target BSs that achieve approximately equal load. Averaging is calculated over a configurable time interval with a default value of 200 frames [8]. Choosing the more suitable BS as a target BS is done through scanning and association activities. The criteria that enlighten the choice may include the link quality in the UL direction. Such handoff monitoring parameters are measured by the MSs and then sent in a form of report to the managed BS, which may deliver them to the radio resource controller (RRC). The RRC, which is generally implemented at the ASN GW or at a BS, controls multiple radio resource agents (RRAs) located at the BS level. RRAs maintain a database of collected radio resource indicators received by the managed MSs or by the respective BSs. In fact, IEEE 802.16d amendments define the receive signal strength indicator (RSSI) and the CINR as two main signal quality indicators that help in assigning BSs and selecting adaptive burst profiles. The specifications also define the mean and the standard deviation because the channel behavior varies in time. These indicators enable the channel quality monitoring and may be augmented to include measurements related to QoS parameters such as the burst error rate. 

It is valuable to note that the RSSI measurements do not require receiver demodulation lock; therefore, they provide reliable channel-strength estimation even at low signal levels. Each BS has to collect the RSSI measurements while each MS shall obtain an RSSI measurement in an implementation-specific fashion. After performing successive RSSI measurements, the MS should derive and update estimates of the mean and the standard deviation of the RSSI and then report them in units of dBm via a REP-RSP message. To obtain such a report, statistics are quantized in 1 dB increments ranging from−40 to −123 dBm while the values outside that range are assigned the closest extreme value within the scale. IEEE 802.16d specifications do not impose how to estimate the RSSI of a single message, but they claim the relative accuracy of a single-signal-strength measurement taken from a single message to be 2 dB with an absolute accuracy of 4 dB. 

Each BS has to collect the CINR measurements while each MS shall obtain a CINR measurement in an implementation-specific fashion. After performing successive CINR measurements, the MS should derive and update estimates of the mean and the standard deviation of the CINR and then report them in units of dB REP-RSP messages. IEEE 802.16d specifications do not impose how to estimate the CINR of a single message, but they claim the relative accuracy of a single signal strength measurement taken from a single message to be 1 dB with an absolute accuracy of 2 dB.

2 Optimization Functions

The handoff process should not decrease the overall performance; therefore, it should implement mechanisms that optimize the delay to support real-time applications such as VoIP and video streaming. Connection dropping should also be reduced. IEEE 802.16e and mobile WiMAX specifications define prescan mechanism to measure the radio connection and select the target BS before handoff execution. Nevertheless, the handoff procedure should include not only layer 2 handoff but also the IP layer handoff for IP-based services. 

Figure 1 below shows that layer 3 handoff, which includes movement detection, IP configuration, and location registration subphases, highly affects the overall handoff delay and requires much more time to be executed than layer 2 handoff. Optimizing such delay is a hot research issue. For instance, the fast handoff procedure proposed proposes to process the configuration of CoA, duplicated address detection (DAD), etc. in advance to reduce the handoff latancy. However, the mobile IP procedure always begins after ending layer two handoff so that the handoff delay is the summation of the layer 2 handoff delay and layer 3 handoff delay. Consequently, it is imperative to consider the correlation between layer 2 and layer 3 within a cross-layer scheme. Analyzed the signaling message flow sequence and the format of both IEEE 802.16e and fast MIPv6 to correlate both handoff procedures and minimize the required signaling overhead. For instance, propose to integrate some layer 3 handoff information with the MOB_HO_IND message and the RNG_REQ as they share semantic characteristics while performing the handoff.

 

Figure 1: The handoff process for IP based services.

Regarding QoS, mobile WiMAX specifications define different classes of services having different requirements in the quality of the traffic delivered to MSs. This quality is generally measured in terms of data integrity, latency, and jitter. The handoff process should be optimized to not highly affect these parameters. For instance, maintaining data integrity during handoff means that the packet loss, duplication, or reordering rates will not be considerably increased while the impact on the DP setup latency/jitter should be minimized. From the QoS point of view, there are controlled handoffs and uncontrolled handoffs. A controlled handoff should respect the following conditions:

• If the handoff is initiated by the MS, the latter should send to its serving BS a list of potential targets.
• Network should base its target selection on the list of potential targets provided by the MS.
• Network should inform the MS with the list of available targets for handoff. If that list is empty, the network will refuse to accept MS handoff. The available targets list should be a subset of the one requested by the MS or reported by it.
• MS should process handoff by moving to one of the provided targets or it should cancel the handoff. The decision is sent via the MOB_HO-IND message.

When any of these conditions is not respected, the handoff is considered uncontrolled and does not provide any QoS guarantees. In the worst case, the MS can connect to the target BS without any indications given to that target BS inducing an unpredictive handoff. To provide data integrity and optimize the handoff process, several mechanisms are provided and classified into two main groups: DP setup mechanisms and DP synchronization mechanisms. DP setup mechanisms refer mainly to buffering and bi-multicasting. Buffering consists in saving the traffic of the services for which data integrity is required at the DP originator or the DP terminator level. 

Nevertheless, the buffering point may change during handoff based on the data integrity mechanism selection. Moreover, buffering may be done only during handoff or for simplicity within the lifetime of a session. On the other hand, multicasting refers to multicasting downstream traffic at the originator endpoint of the DP while bicasting consists in bicasting traffic to the serving element and to only one target. It is worth noticing that bicasting achieves better performances when it is combined with buffering. DP synchronization mechanisms aim at guaranteeing data delivery in different data functions, which buffered the different DPs (serving and target) used to deliver the data during handoff. It is achieved by using either sequence numbers, data retrieving, or Ack window with sequence number disablement. 

A sequence number is attached to each SDU in the ASN DP and then incremented by one every time it is forwarded in the DP. Data retrieving does not require the definition of sequence numbers; the anchor DP function buffers or copies the data during handoff preparation, and when a final target BS is identified, the serving BS will push back all the nonsent packets to anchor/target DF. When Ack window with sequence number disablement is implemented, data storage buffers in anchor DP are released by full or partial ACKs from the serving BS without requiring sequence numbers. Vertical handoff between mobile WiMAX and other wireless networks represents a hot research topic as it requires the implementation of optimization algorithms that are able to decide when to initiate handoff and which access network to choose while minimizing the handoff latency and preserving the security context and the required QoS.