Deployment and Link Planning of Adaptive Coding and Modulation Radio Networks

Operators are faced with the challenge of providing enough transmission capacity for the almost exponential demand in bandwidth from new data services, while constrained by a relatively modest increase in revenue from those services (see Figure 1). This additional bandwidth demand is not an issue for fiber optic transmission systems, which have virtually limitless capacity. However, radio systems are constrained in capacity by available radio frequency (RF) bandwidth and can suffer quality degradations during atmospheric irregularities.

Conversely, fiber is not readily available in the access portion of the network and is too time consuming and costly to provide in a ubiquitous manner. Therefore, microwave radio will continue to be the technology of choice for next-generation access networks, compelling a solution to the problem of constrained bandwidth.

Figure 1 The end of the 'Voice Era' and the increase in demand for data traffic is creating a widening 'Revenue Gap'.

User data throughput can be increased in a fixed RF bandwidth through the use of more complex modulation schemes, improving the bits/Hz efficiency. However, historically, this came at the cost of all the traffic being more susceptible to noise and interference. Thus, there was a trade-off between quality and capacity. This was a one-off decision that had to be made at the design phase, where the choice was either to deploy radios with a low modulation scheme, which could achieve higher availability and performance for a given size of antenna, or have much higher capacity in the same RF bandwidth, but with a lower level of quality for all traffic. Alternatively, to achieve the same Quality of Service (QoS) metrics, extra expense had to be incurred by deploying larger antennas on the radio hops.

ACM Unlocks Additional Capacity

Microwave radio systems have now evolved that can provide significantly more capacity in a fixed RF bandwidth—at the equivalent level of service to fiber. The key technology enabler for this is Adaptive Coding and Modulation (ACM).

ACM radio systems are able to monitor path conditions and switch to lower modulation schemes, creating more available fade margin, when necessary, to extend the primary traffic and critical radio control services by enhancing the availability and performance of the link. ACM systems can thus offer significantly more throughput most of the time, the only trade-off being that additional traffic being carried cannot achieve the same availability and performance as the core traffic.

Transmission networks now have a mix of traditional TDM and Ethernet traffic. Ethernet is becoming more popular in transmission networks, largely due to the efficiencies and scalability for carrying data traffic, which includes voice in the case of VoIP. Using various prioritization methods, Ethernet traffic is easily differentiated to achieve different QoS levels and thus ACM becomes a very attractive solution for reducing the overall transmission bottleneck.

What is ACM?

ACM is a new technology available in radio equipment that allows the coding and modulation schemes to be changed according to path conditions. ACM equipment is designed to ensure these changes do not exceed either transmitter or receiver spectrum regulatory masks. Changes in modulation and thus throughput of data and link uptime can be improved under adverse path conditions. Traffic shaping is done at the radio input to match the offered traffic with available channel capacity. This ensures that prioritized traffic is transmitted at its peak level in order to meet availability and performance requirements. As shown in Figure 2, additional traffic can be carried over the link during non-fading periods, significantly increasing the overall average throughput of the link.

ACM increases the average throughput of data, but at the cost of blocking access to some traffic to the radio path for small periods of time. It is particularly useful when:

• Maximum throughput is more important than reliability;
• Different levels of priority traffic can be identified;
• Licensing conditions insist on high spectrum efficiency modulation schemes, yet extra performance is desired for some critical traffic;
• Licensing conditions allow a simple modulation scheme to operate on the link, but maximum return on the investment in the spectrum allocated is desired;
• Existing links are required to be upgraded to a higher capacity using the same path and antenna sizes.

The term “adaptive” is somewhat misleading, as it creates the impression the links are constantly changing as weather conditions change. Digital radio links operate on a threshold basis, where they are not affected by changing weather conditions until a critical receiver threshold value is exceeded. Any fading activity is thus not seen by users until it breaches this critical threshold. Lower modulation schemes allow this threshold to be breached for less time, thus increasing availability and performance.

However, this margin is not needed most of the time, and the links can operate at this higher throughput with no degradation in quality. The radio monitors the hop conditions, and it only changes the modulation scheme when it is possible that these critical thresholds could be breached. In practice, this happens very seldom, despite the fact that modulation switching is done on a predictive basis, pre-emptively deciding to switch prior to any potential failure event. Adaptive modulation should not be considered as a dynamic optimization technique.

Defining Quality of Service

With the emergence of high speed mobile data, the IP and telecom worlds are converging. The IP world is very familiar with using statistical gains in multiplexing data into asynchronous packets to achieve maximum data throughput. More recently, work has been done to define a specific QoS for “carrier grade” service, which typically refers to technologies that can achieve network availability of 99.999 percent with re-convergence times of 50 ms or less. In the telecom world the focus has traditionally been on providing a dedicated channel, and associated fixed bandwidth and throughput, with clear methodologies to calculate any expected outages or quality impairments on the radio hop.

In a radio-based ACM network, instantaneous throughput can change as link conditions demand. Traffic can be shaped at the radio channel input such that different streams can experience different levels of service. What is important to note is that the background error rate of the traffic itself is unaffected.

QoS allows quality metrics to be defined as part of a customer’s Service Level Agreement (SLA). A top-down design methodology is needed, where the network has to be designed to a specific QoS, which in turn dictates QoS objectives for radio links and individual radio hops.

Quality in a Radio Network

In radio networks, weather irregularities cause additional impairment to background noise in the equipment. Rain outages affect the availability of the link, whereas multipath fading affects the quality of the link, during the period when the link is available.

The ITU standards clearly differentiate between availability standards (defined in ITU-T G.827 and ITU-R F.1703) and performance standards (defined in ITU-T G.821, 826 and 828 and ITU-R F.1668). This is complex to apply to real radio networks, so for short radio links it has been a common industry practice to combine availability and performance in a per hop reliability target of between 99.99 percent to 99.999 percent.

While this has been a useful design simplification for access links, it gets the “right answer” using the “wrong method.” With ACM systems, which can have variable throughput and have different traffic streams experiencing differentiated availability and quality levels, it becomes essential to separate these two aspects again. Otherwise operators will become very confused as to the true performance of the link, and system planners could make incorrect hop design decisions. Availability is the amount of “uptime” of the radio link, and during this period, performance is measured, which affects throughput and transmission quality.

Availability (Rain Outage)

For links greater than 13 GHz, rain is the primary reason for any outage. When the hop fade margin is exceeded by rain of a particular intensity, an outage will occur for the duration of that rain fading event. This could be many minutes and thus affects availability. It can be treated equivalent to a fiber cut, so the fade margin should be chosen to ensure it is a rare event. These outages can be easily quantified if the rain intensity statistics are known or can be estimated.

In ACM links, where each modulation scheme has a different fade margin, the available time for each traffic stream has to be considered separately, as shown in Figure 3. Each differentiated traffic stream will thus have a different level of “uptime,” which is a function of the system gain figure for each modulation scheme.

Performance (Multipath Outage with Throughput Impact)

For links less than 10 GHz, multipath fading—where multiple radio signals with differential delays arrive at the antenna simultaneously—is the primary reason for any “outage.” The atmospheric conditions that cause multipath outages are typically limited to a fading season of some months and occur at specific times of the day (usually several hours at sunrise and sunset).

While unusual for multipath outages to exceed 10 seconds, which, strictly speaking, is not an unavailability outage (i.e., “downtime”), error bursts that often result would mean short blocks of time where quality is so impaired that the service quality is not “acceptable.” This can result in a short break in transmission (i.e., severely errored seconds, loss of frame), which is considered a performance outage. Until the fading depth equals the fade margin, no errors will occur.

Multipath fading that does not cause errors to priority traffic on lower modulation schemes would cause errors to traffic on higher modulation schemes due to the lower fade margin. When the performance levels of the lower- or non-priority traffic is calculated, if the number of fading events is considered too high, the fade margin can be increased with bigger antennas or, better still, space diversity should be used to reduce the outages. Route diversity should not be used to improve performance because multipath fading is a fast event and requires hitless protection switching as provided per space or frequency diversity. The improvement with ACM against multipath fading can be seen in Figure 4.

Interference

Interference is any unwanted signal that has sufficient signal energy within the pass-band of the wanted receiver to affect the demodulated signal quality. For the demodulator to correctly demodulate a wanted signal error-free, the unwanted signal has to be below a minimum signal to noise (SNR) level. The higher the SNR value is, the more complex the modulation scheme. Since the radio hop is required to carry operational traffic all the way down to the threshold of the receiver, interference signals must be considered under faded and unfaded conditions.

In ACM systems, when the modulation scheme changes from a lower order to a higher order, the receiver threshold degrades and the minimum SNR figure required by the demodulator increases. In Europe, where ETSI standards apply, the transmitter power also needs to be backed off to achieve the required linearity using the higher modulation scheme.

Measured from the 256QAM threshold point, the maximum interference level is roughly the same, regardless of modulation scheme. This is because at a lower modulation scheme, although the minimum SNR is relaxed, the reference point is its receiver threshold point, which is much lower than at a higher modulation scheme.

This is significant, as it means that once a frequency is allocated, the maximum interference level allowed will be below the minimum SNR for any modulation scheme. For example, an interfering signal below an unfaded receiver would not cause problems for a 4PSK link, but it would seriously degrade the threshold of a 256QAM radio. However, an interfering signal that high would not be allowed in the first place because it would exceed the SNR requirements of a faded 4PSK hop.

It could be argued that the 4PSK radio would still be more resilient to “rogue” interferers, generated outside of the frequency planning process. However, that negates the benefits of having a licensed frequency. Higher level interferers that exist on a temporary basis under adverse conditions are also addressed by frequency planning and the fact that lower modulation schemes are more robust under this condition—part of the benefit of using ACM in the first place.

While it may appear that a new link could be allowed by the regulator for a radio with a lower modulation scheme, this could breach the SNR requirements of the existing high modulation scheme radio, inhibiting its return to the higher modulation scheme after a fade has occurred. In practice this could not happen.

Traffic Prioritization

Key to achieving maximum benefit with ACM is the ability to prioritize customer traffic. In traditional TDM networks, traffic multiplexed into a PDH-aggregated stream could not be identified with different priority levels because the information would have been bit-interleaved into the aggregate stream. In SDH, each virtual container (VC) could be directed through the network and its performance characteristics measured through the header bytes allocated to the VC. In ATM networks, QoS was an embedded design goal in the standard.

Prioritization of different IP traffic types with differentiated QoS for different types of services was taken into account from the start to allow the router to select a route based on minimum delay, maximum throughput, maximum reliability or lowest cost.

Routing of traffic types can be prioritized by inspecting the packet headers in the switches or routers. It is also possible to define each port of a device with a different priority level. In radio equipment, TDM traffic can be queued according to priority so that both TDM and Ethernet traffic can be differentiated.

ACM Hop and Link Design

ACM requires a change in thinking to traditional radio hop design. Current practices often ignore practical considerations, are not cost optimized and focus on high levels of availability per hop with no consideration of optimizing the throughput in the RF channel. All traffic is treated as being equal despite the fact that the types of traffic have different real QoS requirements and revenue profiles.

In the past, link capacity was fixed and thus the only variables were the fading conditions. To design a radio hop, the availability and performance of the traffic must be calculated. This defines the hardware set-up (i.e., antennas, equipment type, configuration) and software settings (e.g., maximum transmitter power). In the case of ACM, not all traffic experiences the same QoS metrics because some traffic is restricted during anticipated heavy fading to enable changing the modulation scheme and reducing the outage time for the remaining traffic. Radio link design must therefore take into account the improvements made using ACM.

Traditional Methodology

Historically, the approach with radio link design has been to set a ‘hop’ objective and then increase antenna size until the fade margin meets the hop objective. On long hops this often meant that the largest antennas in the manufacturer‘s line were used, which puts huge stresses on towers. ACM can be used to provide extra system gain instead of increasing antenna size, with no infrastructure implications.

ACM Design Methodology

Radio hop design should be designed in a pragmatic way taking into account where the hop is in the network, what type of traffic is carried, the overall network topology and practical considerations of radio and antenna spares, antenna space on the tower and maintenance constraints.

The customer experiences the effects of the holistic design considerations and not just the per-hop fade margin. The ACM design can be accomplished assuming that the traffic outages can be calculated for each prioritized queue, linked to the various modulation schemes deployed, or “virtual pipes.” In the case of a non-prioritized aggregate input, the average outage figure can be computed.

ACM Link Design

In order to design an ACM link, it is important to consider in which of two modes ACM is being used:

Equal priority traffic mode: In this mode the link is designed using the highest modulation scheme and an improvement factor can be calculated for running the link with ACM. This improvement may not result in smaller antennas, but does improve customer experience. The average availability and throughput are improved.

Prioritized traffic mode: In this mode, the link is designed to meet the QoS metrics of the core traffic using the lowest modulation scheme. The performance and availability of the other “virtual pipes” for additional non-priority traffic can be calculated separately. In other words, QoS can be calculated for each priority queue, treating the radio as having n virtual links where n is the number of modulation schemes deployed. Thus, smaller antennas can be used rather than if the whole path had been designed for the maximum modulation scheme.

In either case, the basic QoS figures for the “core” or “all traffic” are tabulated using the conventional outage calculations for availability and performance. The end-to-end outage is calculated using the traditional approach of cascaded hops making up a radio link.

Key Recommendations for Deploying ACM links

A more holistic approach to radio design is needed, where the nature of the traffic in terms of importance (e.g., revenue, cost of failure) and type (e.g., delay sensitive, bandwidth sensitive) should drive design. A top-down design methodology should be assumed, where the radio hop design is a subset of network topology decisions.

In summary, the following are some particular guidelines that should be followed:

• Design to network and link objectives that include system architecture and traffic management considerations, with less emphasis on radio hop design to fixed design limits.
• Ensure the link meets the overall QoS objective, taking network topology into account, and do not iterate antenna sizes until a per hop threshold value is reached.
• Focus less on flat fade margin and large antennas to achieve availability and performance goals.
• Treat availability and performance separately.
• Compare rain outage with annual availability (uptime) objectives for each virtual ACM pipe. The overall average uptime can be calculated for the system.
• Radio link design should be done treating each capacity stream linked to a modulation scheme as a virtual pipe. The radio system gain parameters should be used for the ACM radio that has the built-in ACM coding gains.
• If core traffic can be differentiated, the link should be designed to meet the lowest modulation scheme outage.
• If all traffic is treated equally, the link should be designed for the highest modulation scheme deployed. The improvement in average uptime can be specified.

Conclusion

ACM is a technology available to operators at relatively little additional CAPEX that can significantly optimize the transmission available bandwidth and therefore user throughput of data. The usual trade-off in quality vs. RF bandwidth efficiency is removed by segmenting and prioritizing the input traffic, so that the highest level of availability and performance can be achieved for priority traffic, while achieving the throughput expected using spectrally efficient systems.

ACM provides benefits, but is particularly useful where input traffic can be prioritized with different QoSs. Even where no traffic prioritization is performed, the average availability and performance improves compared to non-ACM, high-modulation systems.

The coding gains through ACM are an excellent alternative to increased gain provided by large antennas, which can enable significant savings in installation and tower costs. In order to design ACM links, QoS must be considered where availability is the uptime period, which is maximized through redundant equipment and network topology, and performance is considered during uptime periods and is optimized to achieve the maximum throughput of data.

Operators are struggling with the threat of insufficient backhaul capacity for new bandwidth hungry data services. Modern ACM-equipped microwave radio systems are now available that provide an ideal solution. Adaptive modulation will be a key technology feature sought by mobile operators who are planning to evolve their transmission networks to all-IP to support next-generation broadband services, enabling them to smoothly and cost effectively introduce high speed Carrier Ethernet transport, with minimal CAPEX demands.