802.16m and WiMAX Release 2.0

Chris Thomas, cat4@cec.wustl.edu (Under guidance of Prof. Raj Jain)


Abstract

Due to continually increasing data rate requirements for mobile networking, the IEEE 802.16 group has put significant effort into the new 802.16m standard. This new standard will serve as a starting point for WiMAX Release 2.0 and brings many benefits over the previous standards used for WiMAX Releases 1.0 and 1.5. This survey paper discusses the advancements that the current form of the 802.16m standard will bring. Specific changes to the PHY and MAC layers are discussed, followed by a summary of information relating to expected performance.


Keywords

WiMAX, Release 2.0, 802.16, 802.16m, PHY, MAC, MIMO, Duplexing, MAC, Backwards Compatibility, Frequency Bands, Bandwidth, ARQ, QoS, Quality of Service, Handover, Femtocells, Relay, Performance, Latency


Table of Contents:




1. Introduction

With the arrival of mobile devices such as smart phones, there is now a demand for wireless networks which both support high data rates and high mobility. WiMAX networks offer a response to this demand. However, with the expectation of constantly increasing data rates comes the need to continually upgrade the standards behind these networks. The next major revision will be WiMAX Release 2.0, which will be based on the 802.16m standard [Yaghoobi09].

1.1 The WiMAX Forum

The goal of the WiMAX forum is similar to the WiFi Alliance. It exists to develop industry standards based on the IEEE 802.16 standard, and then to provide testing and certification of products to ensure interoperability [WiMAX]. Work is in progress on the Release 2.0 standard [Yaghoobi09], but they have not yet published any significant information about the details of implementation. They have made it clear, however, that it will be an implementation of some subset of the 802.16m standard. While the 802.16m standard is not yet finalized at the time of this writing, the 802.16 Task Group m has made substantial information available. As such, the paper deals primarily with 802.16m, and the final implementation details of WiMAX Rel 2.0 may vary slightly. Due to the fact that neither 802.16m nor WiMAX Rel 2.0 is finalized as of the time of this writing, there are no products available on the market.

1.2 Timeline of WiMAX Releases

WiMAX Rel 2.0 will be the third significant release of WiMAX. The first, Rel 1.0, targeted International Telecommunications Union's IMT-2000 standard by implementing a subset of the 802.16e + Cor2 standard [Yaghoobi09]. The final network specification was released in the second quarter of 2007. It employed Time Division Duplexing (TDD), and provided data rates of around 70 Mbps [Yaghoobi09]. Rel 1.5 followed with a final release of the network specification in the fourth quarter of 2008. It employed Frequency Division Duplexing (FDD) and allowed speeds of around 125 Mbps. The Rel 2.0 network specification is expected to be released in 2010 [Yaghoobi09], but an exact date is not yet available. This release will feature data rates around 300 Mbps, employ a combination of FDD and TDD, and is being designed to target the International Telecommunications Union's IMT-Advanced specification.

1.3 Overview of 802.16m

The IMT-Advanced specification which 802.16m is being designed to target is the successor to IMT-2000. The primary improvements include adding support for many new service classes (especially telecommunications services), increased mobility, and better Quality of Service (QoS) guarantees [Bacioccola10].

The stated goals of the 802.16m amendments are “to provide an advanced air interface for operation in licensed bands” [SDD]. It maintains support for legacy devices based on earlier variations of the 802.16 standard, while providing significant performance increases for new devices. Devices, which operate on an 802.16m network, fall into one of several categories. Advanced Mobile Stations (AMS), Advanced Base Stations (ABS), and Advanced Relay Stations (ARS) support 802.16m operation, while R1MSs and R1BSs (Revision 1 mobile and base stations, respectively) are legacy devices.

Back to Table of Contents


2. 802.16m PHY

The PHY layer of a network deals with the physical properties of it. For a wireless network this means the details of the spectrum over which data is transmitted, the details of exactly how that transmission occurs (encoding, antenna techniques, etc), as well as information about the structure of the frames that are sent out.

2.1 Frequency Information

802.16m is designed to operate in a variety of frequency bands in a listened spectrum. The specific bands supported are 450-470 MHz, 698-960 MHz, 1710-2025 MHz, 2110-2200 MHz, 2300-2400 MHz, 2500-2690 MHz, 3400-3600 MHz [SRD]. The 450-470 MHz, 1710-2025 MHz, and 2110-2200 MHz bands are not supported in previous revisions of the 802.16 specification [Yaghoobi09]. Details in regard to channel sizes appear to be unchanged from previous releases and are therefore not discussed here.

2.2 Backwards Compatability

A goal of the 802.16m standard is backwards compatibility with previous releases. This is accomplished by employing Time Division Duplexing (TDD) between a legacy mode and a .16m mode of operation [SDD]. Note that TDD is used to separate these operation modes regardless of how the Duplexing of the connection as a whole is setup [SDD]. This means that if TDD is employed for the general network, each uplink or downlink time slot is divided into legacy and 802.16m sections. Likewise, if FDD is employed for the general network, the uplink and downlink channels are each partitioned into legacy and 802.16m sections.

Time is divided into Time Zones represented as an integer larger than zero that represents several consecutive subframes. A given Time Zone is either an LZone if legacy mode is employed during it, or an MZone if 802.16m mode is employed during it [Yaghoobi09]. This means that generally an R1MS will communicate during the LZone and an AMS will communicate during the MZone. It is possible for an AMS to communicate during the LZone, but it will operate in legacy mode and will not experience any of the benefits of using 802.16m [SDD].

2.3 Frame Structure

The general frame structure is illustrated in Figure 1. Data is organized into a hierarchy of Superframes, Frames, subframes, and OFDM symbols. Superframes last 20 ms and contain four 5ms frames, each of which contains 8 subframes unless the channel size is 7 MHZ in which case frames contain 6 subframes, or 8.75 MHZ, in which case frames contain 7 subframes [SDD]. Subframes fall into one of four categories, three of which are employed as part of 802.16m and a fourth that is included for legacy operation with 802.16-2009 devices operating on 8.75 MHz channels. These types are:
Figure 1 — Frame Structure with Type 1 Subframes
Figure 1 — Frame Structure with Type 1 Subframes

2.4 Duplexing

802.16m supports both FDD and TDD modes. TDD must obviously be half duplex, but FDD operation may be either half or full duplex [SRD]. ABSs must support both half and full duplex modes for AMSs operating with FDD [SRD]. This means that some AMSs may be operating in half duplex FDD mode at the same time that others are operating in full duplex FDD mode on the same ABS, and still more may be operating in TDD mode.

Using FDD has the clear advantage of allowing full duplex mode, but this ability comes at the cost of needing paired channels. While TDD is limited to half duplex mode, it requires only a single channel. TDD also has the advantage that it is very easy to adjust the downlink/uplink ratio by simply altering the amount of time given to each [SRD]. If the circumstances require it, this could even be adjusted to the extreme case of unidirectional communication [SRD].

2.5 MIMO

Multiple-Input-Multiple-Output (MIMO) involves using several antennas, and using processing on the differences in what is received by each antenna to allow for faster data rates [Wikipedia-MIMO]. Multiple Antennas are required for 802.16m operation (though legacy modes may be operational with single antenna setups).

ABSs must support at least 2x2 (two transmitting antennas and two receiving antennas) MIMO, and may have two, four, or eight transmit antennas [SDD]. AMSs must support at least 1x2 MIMO.

There are two forms of MIMO defined in the 802.16m standard. Single User MIMO (SU-MIMO) and Multiple User MIMO (MU-MIMO). These modes employ the multiple antennas in different ways, for MU-MIMO it is primarily to help differentiate between users, while in SU-MIMO it is primarily to increase data rates [SDD].

Back to Table of Contents


3. 802.16m MAC

The Media Access Control (MAC) layer is responsible for encapsulating data and feeding it into the PHY layer for transmission, as well as receiving new data that is verified then passed on to higher layers. Note that higher layers are largely independent of the network they run on and therefore are not dealt with in 802.16m.

3.1 HARQ

HARQ, or Hybrid Automatic Repeat Request is a system used by 802.16m to ensure all packets are transmitted and correctly received. While there are several variations of ARQ, 802.16m uses one primarily based on stop-and-wait [SDD]. This means that when each frame is sent, the sender waits until it received an ACK (acknowledgement) before sending the next frame. Multiple HARQ channels can run in parallel (up to 16), mitigating the performance hit of waiting for an ACK before sending more data [SDD].

Each of these channels has a unique identifier that is determined differently for UL and DL traffic. For DL traffic, it is simply the HARQ Channel ID (ACID). For UL traffic this identifier is a combination of the ACID and the index of the subframe containing the HARQ data [SDD].

Legacy systems used a different version of HARQ, which is supported in 802.16m as a special case. The old version is called Chase Combining, and involves retransmitting exactly the same data in the event that an ACK is not received [Bacioccola10]. 802.16m uses a variant of HARQ known as Incremental Redundancy (IR). IR retransmits the same data with a (potentially) different encoding [Bacioccola10]. The idea behind using different coding is that if there is some interference preventing the data from being received correctly, an encoding with more redundancy or simply a different bit-pattern may either work better, or different portions may be received allowing the incorrectly received data to be combined to create a correct copy [SDD]. Chase Combining is simply the special case where the new encoding is the same as the old one.

3.2 QoS

The MAC layer of an 802.16m network is based on the concept of connections, which are conceptualized as unidirectional data flows (each of which will generally be paired with a data flow in the opposite direction) [Bacioccola10]. Each flow is assigned a four-bit Flow ID (FID). The FID can be combined with a 12-bit Station ID (STID) to generate a network-unique 16-bit identifier for that flow [Bacioccola10]. The separation of FIDs and STIDs is useful due to the fact that in Handovers the FIDs do not need to change. This allows all connections to be very quickly reestablished by simply changing the STID to the new value assigned by the new ABS [Bacioccola10].

There is a downside in that the legacy model allowed many more than the 16 connections per MS limit dictated by the four bit FID. Legacy systems used the full 16-bit connection ID for each connection, which allowed up to 216 connections per station [Bacioccola10]. However, each of these connection IDs had to be reassigned on handover, which created significant overhead.

Each flow in 802.16m may have QoS service parameters [Bacioccola10], which are negotiated between the ABS and AMS when the flow is setup. These parameters are the same as the parameters assigned to connections in legacy systems.

An important part of QoS is bandwidth allocation. The bandwidth request protocol has been reworked for 802.16m. In legacy systems, a five-message request was needed, which specified the bandwidth grant side explicitly each time. In 802.16m there is a shorter 3-message grant request available that will automatically assume some default size and allow two messages to be skipped, thereby lowering latency [Bacioccola10].

3.3 Handover

There are four cases for the handover of an AMS from the currently Serving-BS (S-BS) to a new Target-BS (T-BS). These cases are [SDD]:
1. R1BS → R1BS
2. ABS → R1BS
3. R1BS → ABS
4. ABS → ABS

The first case simply uses the legacy handover protocol, and the second uses the legacy handover protocol in the LZone of the S-ABS [SDD]. The third uses the legacy handover protocol in the LZone of the T-ABS, but after handover is complete if the MS is an AMS, it can transition to the MZone of the ABS it is now associated with [SDD].

The fourth case, in which an AMS is handed over from one ABS to another ABS uses a new protocol comprised of three phases: HO-Initiation, HO-Preparation, and HO-Execution [SDD]. The HO-Initiation phase is only necessary if the Handover is being started by the AMS, and is comprised only of the AMS sending a signal to the S-ABS requesting a handover [SDD].

The HO-Preparation step involves the S-ABS sending authentication and identification information about the AMS to one or more potential T-ABS via the network backbone [SDD]. Each potential T-ABS then performs ranging on the AMS, and the results are returned to the AMS, which then selects which of the potential T-ABSs will actually be targeted [SDD]. Note that this step can be skipped if there is only one potential T-ABS [SDD]. Finally control information is sent from the S-ABS to the T-ABS giving information about whether the handover will be soft or hard, information about the data flows on the AMS (and any other information necessary to optimize the handover), and the time at which the handover will complete and the AMS will lose contact with the S-ABS [SDD].

The HO-Execution Phase involves the AMS and T-ABS going through the network reentry protocol at the appropriate time (designated at the end of the last phase) [SDD]. The AMS disconnects from the S-ABS either before or after this completes (as appropriate). If it occurs afterwards (the handover is soft) then the AMS employs TDD to maintain communication with both ABSs during the handover [SDD]. The Reentry procedure is largely unchanged from earlier releases, except that some extra ranging and channel data may be specified to avoid interference [SDD].

A Handover will always result in some downtime in network communications, even if only to allow for control data to be sent. Table 1 provides a breakdown of the maximum acceptable downtime for different types of handovers between 802.16 BSs.

Table 1 — Handover Downtime [SRD]
Handover TypeMax Downtime (ms)
Intra-Frequency27.5
Inter-FrequencyWithin a spectrum band40
Between spectrum bands60


An 802.16m network can also deal with handovers of some MS to a different Radio Access Technology (RAT). The 802.21 standard for Media Independent Handover is supported to accomplish this[SDD]. The 802.16m standard defines specific handover procedures for several other RATs, including 802.11, GSM/EDGE, UTRA, E-UTRA, 3GPP2, and CDMA2000 [SDD].

3.4 Femtocells

A Femtocell is a short range ABS with very low transmit power compared to a normal (macro) ABS. They are generally set up either for home/small office use or as a hotspot [SDD] and connect to a conventional wired network backbone to provide network connectivity [Bacioccola10]. There are three varieties of Femto ABS [SDD], which primarily relate to policies surrounding Closed Subscriber Groups (CSG). The three varieties are: CSG-Closed, CSG-Open, and OSG (Open Subscriber Group) [SDD]. CSG-Closed only allows members of the CSG to associate with the Femto ABS (and emergency access, as required by law). CSG-Open allows anyone to connect, but gives members of the CSG priority (and will not allow non-CSG members to connect if it would damage QoS to members). It also allows anyone to connect for emergency services [SDD]. An OSG allows anyone to connect with the same priority [SDD].

Femtocells are identified in the SA-Preamble, which both identifies an ABS as a Femto-ABS and identifies the type of Femto-ABS [SDD]. AMSs can maintain a MAC-Address based list of Femto-ABSs for which they are a member of the CSG [SDD].

There are several special cases regarding handover to a Femto ABS. If the ABS is CSG-Closed and the AMS is not in the CSG, handover is only attempted in the case of emergency [SDD]. If the ABS is CSG-Open and the AMS is not in the CSG, handover is only attempted in the case of emergency or if that is the only way to maintain a stable connection [SDD]. The network will attempt to perform load balancing on Femto-ABSs and the overlapping macro ABS. It is actually done by Femto ABSs initiating handovers when their load is too high [SDD].

Because the low range of Femto-ABSs can result in relatively frequent handover between the Femto-ABS and the overlapping macro ABS, Femto ABSs and AMSs associated with them may cache significant amounts of information to speed up handovers [SDD]. While this is also possible strictly between macro AMSs, it is not worth it due to the large range making handovers rare.

Femto ABSs have a low duty mode similar to power-save mode in a MS [SDD]. The goal of this mode is not to conserve power (as they will generally not be reliant on batteries), but rather to avoid excessive noise in the frequency band that could cause interference with neighboring Femto ABSs or communication between an AMS and a macro ABS. A Femto-ABS can only enter this low duty mode if no AMSs are associated with it [SDD].

Back to Table of Contents


4. Performance

As the primary goal of 802.16m is to increase network performance, a substantial amount of work revolved around specifying performance minimums and expectations. In this section the effect of several factors on performance is discussed. Unless otherwise specified, all performance metrics given in this section are under ideal conditions. Generally performance is expected to increase by at least a factor of two for both uploads and downloads under the same conditions as the previous release.

4.1 MIMO Data Rates

The MIMO configuration used will have a significant affect on performance. Table 2 shows the minimum-expected and target performance of several MIMO configurations. Note that it is far from an exhaustive listing of possible configurations, but merely serves to demonstrate the effect that MIMO can have.

Table 2 — MIMO Performance [SRD]
Requirement TypeDirectionMIMO Config.Peak Rate (bps/Hz)
BaselineUplink2x28.0
Downlink1x22.8
TargetUplink4x415.0
Downlink2x46.75

4.2 Geographic Factors

There are two geographic factors that can have a serious effect on data rates [SRD]. The first is distance between the AMS and the ABS, and the second is travel speed for the AMS. Due to the fact that landscape features will have a significant effect on either of these factors, specific data rates are not specified, only general ranges over which service decays. Table 3 summarizes the performance effect specified in [SRD]. Note that in Table 3 the performance degradation level specified in the middle column is experience when either the distance from the AMS to the ABS is as indicated or the travel speed of the AMS is as indicated. Having both factors in the same range would cause both to performance to degrade faster.

Table 3 — Geographic Factors [SRD]
Speed of AMS(km/hr)Performance DegradationDistance from AMS to ABS (km)
< 10 (walking)None0-5
10-120 (driving)Gradual Decline5-30
120+May not be able to maintain connection30+


In regard to specific speeds at some distance from the ABS, as explained above it is not possible to give numbers that will hold across deployments. However data rates are specified in Table 4 for average users and users at the edge of a cell. Note that the location of the cell edge will vary depending on factors of the geography, etc. of deployment location, but these numbers should be accurate as a baseline wherever that edge is.

Table 4 — Data Rates Near Cell Edge [SRD]
MetricDL (bps/Hz)UL (bps/Hz)
Average User Throughput0.260.13
Cell Edge User Throughput0.090.05

4.3 Latency

Latency should be improved for all tasks when compared to the previous release. The User Plane Latency is defined as the time it takes a packet to travel from the MAC layer of an AMS to the MAC layer of an ABS (or vice versa). For 802.16m, this is specified to be at most 10ms [SRD]. The Control Plane Latency is defined as the time it takes an AMS to go from an idle power-save state to fully active and communicating. This is defined to be at most 100ms in 802.16m [SRD].


5. Summary

802.16m provides a number of improvements over legacy systems while maintaining backwards compatibility. The WiMAX forum is currently in the process of designing the new WiMAX Rel 2.0 based on the 802.16m standard. The PHY layer has been changed in several ways. More frequency bands are usable than in previous revisions. The Frame structure has been updated to provide more variations on subframe structure. Better support for MIMO and more Duplexing options were added (now both TDD and FDD are supported). A system was also put in place to ensure backwards compatibility with older releases.

The MAC layer also has a number of changes. The HARQ system was redesigned to use incremental redundancy instead of chase combining. Significant changes to how connections are processed (and QoS updates to work with the new format) were made with the introduction of the concept of data flows. The Handover procedure received a significant amount of work, especially in defining handover to and from legacy BSs. Support for Femtocells was made significantly more robust.

All of these changes resulted in significant improvements to performance, with a throughput gain of at least 2x and a decrease in latency in all cases.

Back to Table of Contents


6. References

(In order of importance)

  1. [SDD] Srinivasan, Roshni (ed), Hamiti, Shkumbin (ed), "IEEE 802.16m System Description Document (SDD)," IEEE 802.16 Task Group m, September 2009, http://www.ieee802.org/16/tgm/docs/80216m-09_0034r2.zip

  2. [SRD] Cudak, Mark (ed), "IEEE 802.16m System Requirements," IEEE 802.16 Task Group m, January 2010, http://ieee802.org/16/tgm/docs/80216m-07_002r10.pdf

  3. [Yaghoobi09] Yaghoobi, Hassan, "Mobile WiMAX Update and IEEE 802.16m," IEEE, March 2009, http://ieeetmc.net/r6/scv/sps/WiMAX_Update_802-16m.pdf

  4. [Bacioccola10] A. Bacioccola, C. Cicconetti, C. Eklund, L. Lenzini, Z. Li, E. Mingozzi, "IEEE 802.16: History, Status and Future Trends", Computer Communications, Volume 33, Issue 2, 15 February 2010, Pages 113-123, http://www.sciencedirect.com/science/article/B6TYP-4XNF43R-1/2/f0a4cc67a688adb985355e0ea97d642f

  5. [Wikipedia-MIMO] "MIMO," Wikipedia, March 2010, http://en.wikipedia.org/wiki/MIMO

  6. [WiMAX] "About the WiMAX Forum," WiMAX Forum, June 2001, http://www.wimaxforum.org/about

Back to Table of Contents


7. Acronym List

3GPP2 3rd Generation Project Partnership 2
ABS Advance Base Station
ACK Acknowledgement
AMS Advanced Mobile Station
CDMA2000 Code Division Multiple Access 2000
CSG Closed Subscriber Group
EDGE Enhanced Data rates for GSM Evolution
E-UTRA European-UMTS Terrestrial Radio Access
FDD Frequency Division Duplexing
FID Flow ID
GSM Global System for Mobile Communications
HARQ Hybrid Automatic Repeat Request
IEEE Institute of Electrical and Electronics Engineers
MAC Media Access Control (Layer)
MIMO Multiple-In-Multiple-Out
OSG Open Subscriber Group
PHY Physical (Layer)
QoS Quality of Service
RAT Radio Access Technology
STID Station ID
TDD Time Division Duplexing
ULTRA UMTS Terrestrial Radio Access
WiFi Wireless Fidelity
WiMAX Worldwide Interoperability for Microwave Access


Back to Table of Contents


Last Modified on April 21, 2010
Note: This paper is available online at: http://www.cse.wustl.edu/~jain/cse574-10/index.html