To address the burgeoning market opportunity for low-cost mobile voice and broadband data services, many new entrants are hoping to steal even a small share of the market from the behemoth wireless carriers.

New-entrant service provider partners, such as competitive local exchange carriers (CLECs), satellite companies, even foreign incumbent local exchange carriers are in the hunt to deliver mobile VoIP/data services to the market on a nationwide or selected market basis.

Interestingly, these providers are taking a different angle to their business plan. They understand something very simple. While mobile data services represent an important, high-growth area, mobile voice remains and will continue to remain the major revenue earner for mobile operators. Note the following market facts:

•   According to telecom research firm Ovum, voice still
    accounts for nearly 70% of cellular industry revenues -- and
    as late as 2014, its share of revenues won’t dip below 60% in
    any region of the world.
•   The GSM Association estimates that of the 4B mobile users
    worldwide, roughly 90% are voice only users.
•   ABI Research estimates that 2010 worldwide mobile voice
    revenues will be $580B.
•   This is contrasted with ABI Research’s estimate that 2009
    worldwide mobile data revenue was approximately $169B
    with roughly half of that being messaging related.

Currently, mobile broadband data revenues are less than 20% of those of mobile voice. Even using bullish industry assumptions for mobile broadband data growth, it is likely to take 9-10 years before mobile broadband data revenues are on parity with mobile voice revenues. Thus, networks that allow service providers to offer mass-scale mobile voice in addition to mobile broadband data as they see fit stand to significantly improve network profitability.

In this article, we seek to compare a new technology called xMax with IEEE 802.16e (aka WiMAX) to determine which of the two technologies, xMax or WiMAX, provides a better return on investment (ROI) for new entrant service providers.

This tutorial covers brief technical overviews of both technologies, highlighting physical (PHY) and Media Access Control (MAC) layer differences and network architecture similarities.

xMax Network Architecture
The primary consideration in the design of the xMax system is the desire to provide robust, scalable, and full-featured voice and data services to mobile subscribers at a fraction of the cost of traditional approaches. A reference model form of the resulting Internet protocol (IP) centric network architecture is shown in Figure 1. As the diagram indicates, the network architecture includes the following elements:

Figure 1. xMax Network Architecture Reference Model.

•   Air-interface designed for operation in unlicensed as well as licensed bands.
•   Basestation (BSN), which provides radio to network access to handsets, and Access
    Network Gateway (ANG), which provides call process and IP packet delivery
    services. The xMax Basestation Node (BSN) is a 3 sector, 18 channel VOIP
    basestation transceiver device. The BSN channelizes the 902-928 MHz band into 18
    discrete channels, which are used only when there is traffic to mobile devices that
    are registered with a particular channel.
•   Radio spectrum utilization is highly efficient in that the system occupies only that
    spectrum which is necessary for individual data streams. The bandwidth of each
    channel is 1.44MHz with sufficient guard band between channels.
•   Technology agnostic backhaul links from BSN sites and the ANG (Fiber, Metro
    Ethernet, PTP Wireless, etc.).

While the BSN is conventional in both architecture and functionality, the ANG is a novel piece of wireless infrastructure equipment that consists of the following:
•   Ethernet Switch which aggregates BSN links.
•   Firewall which provides private to public network address translation (NAT)
    services SIP.
•   Proxy Server which supports SIP call control, SIP message compression, and E911
    services.
•   Proxy DHCP Server which is used for IP address services.
•   Network Monitor which is responsible for end-to-end network management and
    monitoring services.

The architecture further supports the service deployment cost objectives by leveraging commercial off the shelf (COTS) voice over IP (VoIP) equipment, software and services, shown as VoIP core in Figure 1. The VoIP core contains the following elements:
•   SIP Proxy Server which provides traditional SIP call control services.
•   Media Gateway – which provides media transcoding between IP and PSTN networks
     and is responsible for subscriber accounting/billing, PSTN call termination, “Direct
     Inward Dialing” (DID) phone number maintenance, voicemail services, and inter-
     network call signal routing, among others.

We note that for most deployments it is expected that the VoIP core will be operated and managed by a third party partner so the CapEx and maintenance costs associated with this equipment is considered out of scope.

xMax Physical Layer and MAC
xMax is a single-carrier system with forward error-correction (FEC), interleaver, symbol mapper, and pulse shaping filter. The baseline system supports 1.3 Mbps aggregate per channel with a single-order modulation and all 18 RF channels in use. The xMax system is capable of providing 23 Mbps aggregate. Higher rates are possible with higher-order modulation and xG is currently working on adaptive modulation extensions to the system which has the potential to substantially increase the system’s data capability.

Those familiar with wireless system physical layer design will note that the xMax is a conventional single carrier system.

The design of xMax MAC (aka xMAC) protocol was driven by the observation that existing MAC protocols (e.g., IEEE 802.11, WiMAX, HSDPA) do not meet latency and scalability requirements for voice services. As an example, the downlink latency associated with the IEEE 802.11 MAC becomes unacceptable for > 12 concurrent calls. Similarly, the downlink latency associated with the WiMAX MAC with basic scheduler becomes unacceptable for > 17 calls.

In designing xMAC, the xG engineering team focused on meeting the following requirements:
•   Deterministic latency (30msec) independent of number
    of concurrent calls.
•   Adaptable to changes in voice codec.
•   Scalable to 300+ concurrent calls.
•   Supports several thousand registrations.
•   Support of IP traffic (i.e., QoS, Admission Control).

The resulting xMAC is a heterogeneous MAC protocol wherein timeslot based access (TBA) is used for voice and broadband data sessions, and contention based access (CBA) is employed for signaling and short message applications (e.g., SMS).

Finally we note that in order to meet the objective of providing low-cost mobile voice and broadband data services, the xMax RAN solution has been designed around commonly used and open Internet protocols (e.g., SIP, RTP, UDP, IP, etc.), and designed to operate in both unlicensed spectrum, such as the 902-928 MHz unlicensed ISM band, and licensed spectrum. As a result of these design considerations xMax includes responsive interference “detect and avoid” (DAA) techniques capable of 33 channel changes (handoffs)/sec that are suitable to combat any in-band interference encountered in the unlicensed spectrum, and extends the SIP and RTP protocols to the wireless domain.

WiMAX IEEE 802.16e Network Architecture
As shown in Figure 2, WiMAX IEEE 802.16e also has an open IP-based architecture. The network architecture includes the following elements:

Figure 2. WiMAX Network Architecture Reference Model (Source: Intel Presentation).

• Air-interface designed for operation in licensed 2.5 GHz and 3.5 GHz bands as well as the 5.8 GHz unlicensed band,
• Basestation (BS), which provides radio to network access to handsets, and Access Service Network Gateway (ASN), which provides handover, authentication, radio resource management, IP address allocation, etc. The WiMAX Basestation (BS) is typically a 3 sector unit with support for 3.5, 5, 7, 10, or 20 MHz channel widths,
• Technology agnostic backhaul links from BS sites and the ASN (Fiber, Metro Ethernet, PTP Wireless, etc.),
• Core Services Network (CSN) which contains the home agent (HA), authentication/authorization/accounting (AAA) server, and other elements such as the VoIP core used in the xMax network.

Similar to the xMax network, the BS and ASN are considered to be managed and operated by the network access provider (NAP) while the CSN is considered to be operated by a network service provider and therefore out of scope for this analysis.

WiMAX IEEE 802.16e Physical Layer and MAC
From a physical layer perspective, one significant difference between xMax and WiMAX IEEE 802.16e is the fact that WiMAX is based on orthogonal frequency division multiplexing (OFDM) which has much higher spectral efficiency and better resistance to multipath fading than does the single carrier modulation technology used by xMax. WiMAX extends classic OFDM in the sense that it is scalable and allows the use bandwidths ranging from 1.25 MHz to 20 MHz. Support for multiple input multiple output (MIMO) technology further enhances the potential spectral efficiency advantage of WiMAX over xMax.

The WiMAX MAC contains a scheduling algorithm that enables the BS to control quality of service (QoS) by balancing time slot assignments among the handsets and their respective application needs. In principle WiMAX supports the QoS requirements for different types of services including Unsolicited Grant Services (UGS) for services such as VoIP without silence suppression, real-time Polling Services (rtPS) for services such as MPEG video or VoIP with silence suppression, non real-time Polling Services (nrtPS), and Best Effort Services (BE) for regular Internet browsing, etc.

The basic MAC scheduler for WiMAX has been shown to lead to an inordinate amount of overhead. So-In, et. al. "Capacity Evaluation of IEEE 802.16e Mobile WiMax" has calculated that the overhead is such that WiMAX can only support 4-7 simultaneous voice sessions per MHz of spectrum with today's scheduler. That number is shown to be improved by up to a factor of 10 in enhancements being proposed in future versions of WiMAX. Since the MAC scheduler could, in theory, be improved with current IEEE 802.16e systems, in this analysis we use a voice capacity number closer to what could be theoretically achieved rather than what is being demonstrated in systems today.

Maxing-Out
xMax is uniquely positioned for low-cost operation in the 902-928 MHz unlicensed band. Its inclusion of rapid detect and avoid technology together with its use of wireless extensions of the SIP and RTP protocols enable it to operate robustly in the presence of the types of interferers inherent to the 902-928 MHz unlicensed band. In several ways, the 902-928 MHz band is advantageous to new-entrant operators not only because it is license-free but also because RF propagation and building penetration at 900 MHz is significantly better than at 2.5 GHz -- for the operator this means significantly less site acquisition and equipment cost (i.e., much lower CapEx).

WiMAX is a well-designed wireless system and its multi-carrier OFDM framework gives it an inherent raw data capacity advantage at a basestation level over xMax. WiMAX, unfortunately, has two issues that largely blunt this capacity advantage:

Issue 1. As pointed out earlier, the baseline WiMAX MAC scheduler is not able to meet QoS requirements for voice at high load and as a result uses significantly more effective bandwidth per voice call than does xMax.

Issue 2. WiMAX is designed for operation in either the 2.5 GHz band, 3.5 GHz band, or the unlicensed 5.8 GHz band. While these bands are "cleaner" from an in-band interferer perspective than the 902-928 MHz unlicensed band used by xMax, the RF propagation and penetration characteristics of these band are significantly worse and, as a result, the new-entrant operator incurs significantly more network rollout costs than is the case with xMax. One could entertain the notion of deploying WiMAX in the 902-928 MHz band but it is highly unlikely that a WiMAX network will be able to deliver quality voice and data services in this band as they have not been designed with this in band mind, and it is not clear how they will perform in interference rich environments. It's even unclear how WiMAX will perform in the unlicensed 5.8 GHz band where wideband interferers, which can be more problematic than narrowband frequency hoppers, are prevalent.

In this article we compared xMax with IEEE 802.16e (aka WiMAX) to so the network differences can be better understood by providers looking for a voice-friendly solution to their wireless objectives.

New entrant service providers who anticipate higher revenue from voice services than broadband data services in the near-term could be well served by utilizing xMax. The ROI advantage of xMax stems partially from the use of unlicensed spectrum and partially from its explicit support for voice services. Its 902-928 MHz RAN architecture better matches the voice/data capacity per coverage area requirements that some service providers are likely to encounter as new entries in mature markets.

Endnotes:
ABI Research: www.abiresearch.com
GSM Association: www.gsmworld.com
Ovum: www.ovum.com

 About the Author
A lifelong inventor, Joseph Bobier invented xG’s Technology, Inc.’s core technology. He also holds several additional patents for wireless technology. Joseph Bobier co-founded xG Technology, Inc. with Rick Mooers and Roger Branton in August 2002. For more information, visit www.xgtechnology.com. This article is adapted from the white paper titled “Stacking xMax Against WiMax” by xG Technology, Inc., March 2010.

What’s your take on this subject? Leave a comment and get the conversation going.