Backhaul for WiMAX & LTE: High-bandwidth Ethernet Radio Systems
From the Microwave Journal November 2008 Supplement
With the current deployment of high-bandwidth access networks happening at an ever-increasing pace, the need for efficient, high-bandwidth backhaul infrastructure becomes more and more relevant. In fact, the construction of these new access networks depends upon the backhaul network for both technical and financial performance. Today’s cellular networks typically deploy leased T1 or E1 lines to get user traffic to/from base station sites in a given metropolitan area network.
It is not unusual for a large metro city to have many hundreds or even thousands of base station sites. As WiMAX and other high-bandwidth mobile access technologies like LTE grow, deployment bandwidths on the order of 40 to 100 Mbps per base station are envisioned. With this large increase in bandwidth demand and the connectionless nature of the traffic comes the need for a highly capable, wireless metro Ethernet backhaul solution, one that can be deployed faster and more cost effectively than its predecessors or its optical cable counterparts.
This article investigates the application of high-bandwidth radio technology to the creation of the next generation metro backhaul networks. Typical network topologies and operating attributes are highlighted with emphasis on how these impact the underlying radio technology implementations.
What is Driving the Bandwidth Demands in Mobile Networks?
Figure 1 Voice and data trends in mobile networks (source: International Wireless Packaging Consortium [IWPC], Milan 2008).
Historically, bandwidth demand in mobile networks has been driven by voice services and services related to voice. This translated into steady growth for many years. However, there is now a large push for mobile data services. The industry has now seen that data-based services represent a larger portion of overall network traffic (see Figure 1).
Data traffic has a different growth attribute than voice, primarily because general voice calling is dependant upon a ‘human’ application, which has not significantly changed. Our patterns of phone conversation have not changed much over time. Data sessions are largely dependant on the latest applications or services being consumed. For example, a low data-rate e-mail retrieval session can become a very high-bandwidth file download session. Also shown in the figure is the trend toward Ethernet backhaul. This is no surprise since the selected transport technology generally performs most efficiently when it is transporting information already “packaged” in the same transport format. In this case, as packet-based data traffic dominates, it is naturally more efficiently transported in Ethernet form.
Figure 2 Applications of various access technologies (source: IWPC/Alcatel, Milan 2008).
In response to this, the industry has evolved access technologies to enable more access data bandwidth to be available to end-users. Figure 2 shows this trend. Additionally, it also shows the user-demanded access bandwidths and the backhaul technologies being applied. There are various technology labels along the plotted curve. These represent differing technology performance benchmarks associated with different technology streams over time. The vertical dotted line indicates the current position of the evolutions. For instance:
- The CDMA stream evolves to EVDO_rev0, then to EVDO_revA, then to EVDO_revB, then to EVDO_revC;
- The GSM stream evolves to HSDPA, then HSUPA, then LTE;
- The WiMAX stream evolves from 802.16d (fixed/nomadic) to 802.16e (full mobile)
In general, these technologies can be seen as competitors since various operators in a given service area would deploy one or the other. But they are similar in that they are all trying to deliver the required [high-bandwidth] access services. Further, they also have a common need for escalation in their demands on high-bandwidth, Ethernet-based metro backhaul networks. This latter element, the needed backhaul, is a vital but often unnoticed network segment. As obvious as it may seem, the concept of needing a large increase in backhaul bandwidth to support the corresponding large increase in access bandwidth is a much-overlooked, yet critical design element.
Figure 3 Global trends in backhaul implementation/transport technology (source: Infonetics Research - Mobile Backhaul Equipment 2007).
Wireless implementations of metro backhaul have long dominated in Europe. In North America, however, more TDM copper backhaul has been historically employed primarily as a result of low cost ILEC T1 TDM circuits available through US unbundling regulations. The onset of high-bandwidth backhaul demands is incompatible with these T1 circuits; wireless is therefore increasingly seen as a more optimal backhaul technology in this geographic region as well. Cumulatively, as shown in Figure 3, there is a positive trend toward wireless Ethernet technology use, while PDH/SDH are declining. Interestingly enough, these trends are consistent with the bandwidth evolution overview. The connection between the two is that the high-bandwidth access technologies are increasingly relying on high-bandwidth Ethernet backhaul.
Typical Wireless Metro Ethernet Backhaul Networks
Figure 4 Metro backhaul network sub-circuit topologies (source: Base Station, February 2008).
With the increasing focus on high-bandwidth Ethernet backhaul networks to support the newer mobile access technologies comes an ability to harvest a series of inherent networking benefits that Ethernet can offer. Ethernet’s cost effectiveness and connectionless nature make it highly suitable for distributed, self-hardened network topologies. The designer is free to design conventional, daisy-chained, ring or constrained mesh backhaul topologies (see Figure 4), all of which are readily addressable with Ethernet as a transport structure. In addition to the cost effectiveness of Ethernet-based implementations, the ring and constrained mesh topologies offer a number of technical advantages, namely:
- Use of angle diversity to increase radio path availability. This can allow for the use of smaller antennas and/or can allow longer paths;
- Use of geographic-separation diversity to increase radio path availability; this can allow for the use of smaller antennas and/or can allow longer paths;
- Ability to harden “N” network elements within a given sub-circuit with “N+1” cost incremental
The constrained mesh can also be considered as a future bandwidth expansion to the ring. Starting with a ring structure, a “crossbar” radio link can be added in the future to effectively double the overall throughput of the sub-circuit. When this is combined with software-based, highly-scalable radio link throughput capability, individual sub-circuits can be readily engineered to scale from a few 10’s Mbps [full duplex, Committed Information Rate (CIR)] to many Gbps. Covering a metro service area often requires the replication of the backhaul sub-circuit topology in order to provide connectivity to many end BTS sites in the network. It is not uncommon for mid/large metro networks to have hundreds or even thousands of base station sites. In the future, networks using micro- or pico-cellular technologies, higher throughputs, or higher access-layer operating frequencies will drive the base station count up further, possibly by a factor of 10 or more.
Top Level Requirements Placed on the Underlying Radio System Building Blocks
With the drive toward advanced wireless metro Ethernet backhaul networks comes a series of new demands on the radio systems. Historically, TDM radio systems with relatively low throughput could be implemented with simple modulation schemes and low-fidelity transmitters and receivers. In addition, the availability of licensable spectrum at lower frequencies allowed these radio systems to attain long range without unattainable transmit power levels.
In today’s deployment arena, spectrum scarcity and link deployment density are driving deployment toward higher frequencies where it is possible for operators to acquire radio channels from the local regulators. The advantage of this is that larger RF channels are often available (which helps to enable higher data rates over the air). However, the higher rain-related propagation losses at higher operating frequencies create disadvantages such as:
- Need for more RF transmit power
- Need for increased transmitter linearity
- Need for better receiver sensitivity
These disadvantages are partially offset by the ability to attain higher antenna gains [for a given antenna size] at higher operating frequencies.
In terms of operating data rates, the need to attain very high throughputs [on the order of 200 to 500 Mbps Full Duplex (FD) CIR] drives the need for high-order modulations. When combined with high-frequency operation of the radio systems, this creates extremely difficult challenges in:
- Attaining ultra-low phase noise local oscillator operation at high frequencies
- Low noise figure
- Low contributed noise (ultra-"quiet" electronics)
- High transmitter and receiver linearity
- Low AM/AM & AM/PM distortion transmitters
These requirements become significantly more aggravated when the radio systems are deployed in narrow channels (such as 14 versus 56 MHz) while simultaneously being licensed in high operating frequencies (such as 26 versus 13 GHz). This scenario is typical in European markets where operators, “rewarded” for high spectral efficiency, are pushed to narrower channels to minimize costs and ease constraints to access channels and are driven to higher operating frequencies due to congestion. Historically, the need for low delay operation was driven by voice-centric services. In the era of data-centric services, one line of thinking has been that delay is not a concern since data can be highly delay-tolerant. However, there are numerous data applications that are delay and delay-variability sensitive, including:
- Live interactive video conferencing
- Mobile networking (hand-offs)
- Live gaming
- Online navigation aids
- T1 or E1 over Ethernet
The result is that delay and delay variability are highly relevant and important operating parameters. In a typical network, a given backhaul sub-circuit will be composed of 8 to 10 radio “hops” to the nearest fiber. Typically 1 to 2 ms of delay will be allocated within the network. This translates into a necessary delay of approximately 100 to 200 µs per radio hop. When considering mesh and ring topologies, it is possible to also employ the sub-circuit topology to help control delay and delay variability. Delay variability of ±0.5 to 1 ms is readily achievable provided ultra-low-latency radios are employed in an appropriate topology.
Typical Constrained Mesh Metro Backhaul Network Architecture
Figure 5 Typical metro backhaul network construct using ring/mesh sub-circuit building blocks (source: IWPC, Washington 2007).
Figure 6 Typical link distances in a large city (source: IWPC, Washington 2007).
When the ring/mesh topology is employed to deliver metro backhaul coverage into a mobile access network, the network may look as shown in Figure 5. This figure shows the utilization of the ring/mesh sub-circuit building blocks in the creation of a network able to deliver backhaul coverage across a large metropolitan service area. The use of high-performance Ethernet radio systems to realize each microwave/wireless link in the backhaul network enables the network to effectively extend the reach of the metro fiber out to each distributed base station site. To do this, a high system gain is required from the radio systems in order to attain high availability, while using the smallest possible antenna sizes. Figure 6 shows a typical distribution of link ranges from a network design in a large US city. Typical distances are 4 to 5 km, but the possible range of distribution is roughly 1 to 8 km. Note also that the distribution is not “well behaved” (that is normally distributed).
The system gain is a radio parameter, which is generally defined as the difference (in dB) between the average power in the transmitted signal at the antenna-transmitter interface and the Minimum Detectable/Processable Signal (MDS) in the receiver, measured at the antenna-receiver interface. It does not normally include antenna gain. High performance, high-frequency radio systems require as much system gain as possible. Since MDS is largely defined by parameters that are not easy to improve upon, MDS=10 log (KTB)+SNR @ BER=10E–6+NF, the power in the transmitted signal remains the obvious vehicle for performance improvements. Low cost, high P1dB, high IM Intercept, low distortion transmitter MMICs are required to address this performance element.
Next generation access networks are driving wireless backhaul demands to deliver cost and performance that has not previously been available. A number of key system performance attributes fall directly or indirectly to the performance of the microwave front-end designs.
Erik Boch received his MS degree in electrical engineering from Carleton University, Ottawa, and is a registered professional engineer. He has held senior engineering or technical management positions at a number of communications and aerospace companies, including Litton Systems, ComDev, Lockheed Martin and Alcatel Networks (formerly Newbridge). While at Alcatel, he was AVP of the Wireless Systems Group and was involved in various aspects of microwave and millimeter-wave subsystem and system design for more than 22 years. He led the R&D team at Alcatel that introduced the first ATM-based Fixed Wireless Access System in the industry.