Smart Carrier Aggregator (SCA)

/Smart Carrier Aggregator (SCA)
Smart Carrier Aggregator (SCA)2018-09-06T17:27:17-04:00

SCA is an algorithm hosted in our aftermarket box. This box sits in between an LTE cellular site’s baseband unit and its remote radiohead, connecting via its CPRI interface.


PlusN’s Smart Carrier Aggregator (SCA) optimally combines input signals in between the Baseband Unit (BBU) and the Remote Radiohead Unit (RRU) in downlink CA.

Carrier aggregation is the combination of spectrum that is not necessarily contiguous. When aggregated, the spectrum is treated as if it is one large bloc of more manageable spectrum instead of multiple Carrier Components (CCs). Carrier aggregation realizes gains in data rates, but fails to reach its full potential because it creates power inefficiencies that limit the cellular site in range (for rural LTE) or forces the use of techniques that trade off distortion in the signal to address the power inefficiency (for urban LTE). This distortion limits the use of constellation, which affects aggregate data capacity.

This power efficiency arises from combining waveforms originally each directed to its own CC on downlink into one waveform. Commercial wireless has strict latency requirements. Data in cellular sites moves at the rate of one framework per millisecond, creating a challenging environment for optimally combining these waveforms. Carrier aggregation launched with the arbitrary combination of these waveforms. Each waveforms, even in small duration, will include multiple frames that create local maximums of power requirements in the power amplifier. Arbitrary combination of waveforms can result in frames in the combined waveform such that this frame in more than one of the original waveforms created a local maximum in power requirements in the power amplifier. This new peak of power requirements is a super peak larger than anything that would have resulted without carrier aggregation. Furthermore, at peak demand hour, any given few minutes will see multiple frames that line up these peaks from original waveforms.

PlusN’s SCA – while meeting commercial wireless latency requirements – shifts the phases of input waveforms so as to avoid combining frames from multiple waveforms that result in peak power requirements at the power amplifier. The SCA-combined waveform has a much lower Peak-to-Average Power Ratio (PAPR) than a waveform combined arbitrarily.

High PAPR limits the use of power in the cellular site, which limits the range of the transmission. A workaround, especially in densely populated urban environments, is to reduce the PAPR to tenable thresholds through Crest Factor Reduction (CFR) or tone injection methods that introduce Error Vector Magnitude (EVM). The EVM restricts the use of constellation for data transmission – meaning that the capacity is lower for that initial increase in PAPR – but the net gains from carrier aggregation are still worth it. In the first example – rural environments in which range is compromised, inserting SCA into the cellular site immediately increases the range. This example actually is more similar to 5G urban examples than the LTE urban is because of the emphasis on range. In the LTE urban example, the SCA-combined waveform will have a lower PAPR than an arbitrarily combined waveform, but the PAPR still will be higher than required thresholds. The box installed between the baseband unit and the remote radiohead is equipped with a tone injection algorithm that will reduce the PAPR to a configured level. This tone injection still will introduce some EVM, but it a substantially lower amount than that from arbitrarily combining the waveforms for carrier aggregation and then relying on CFR to reduce the PAPR. In the example below – 60 MHz of aggregated bandwidth – the difference in EVM in an example with arbitrarily waveform combination and CFR as compared to an example with SCA and tone injection affects whether the cellular site can use 64-QAM or 32-QAM.


The default PAPR-EVM tradeoff curve is below in grey for the specific example of aggregating 3 CCs each of 20 MHz. The PAPR measures the efficiency of a cellular site. A high PAPR speaks to a host of challenges at peak demand hour: range and heat dissipation. The EVM measured at the receiver is the error when attempting to interpret data in a constellation. There is a tolerable amount of EVM for each constellation, decreasing with higher constellations. The example below assumes that we want to minimize EVM specifically for a 7-dB PAPR tolerance, which enables upgrading the constellation if the EVM decreases sufficiently.

On the gray line – an example without SCA – the network begins with a signal that has a 13.0 dB PAPR from a waveform combined arbitrarily before introducing any EVM. To reduce the PAPR to the 7.0 dB tolerance, the cellular site uses CFR, introducing additional error with an EVM of more than 6.0%.


The grey line shows the PAPR-EVM tradeoff curve with CFR. In order to get from an initial PAPR of 13.0 dB to a tolerable PAPR of 7.0 dB, the signal assumes more than 6.0% EVM in a “clipping” process. The blue line shows the PAPR-EVM tradeoff curve if the signals are combined with SCA instead of an arbitrary Power Combiner.

  1. SCA: PAPR decreases from 13.0 to 9.5 dB (with no EVM increase).
  2. Tone Injection: PAPR decreases from 9.5 to 7.0 dB, EVM increases by 2.5%.

The blue line above shows a version of a similar tradeoff curve with SCA combining the individual waveforms. By optimally combining the waveforms, SCA reduces the PAPR to 9.5 dB before introducing additional error (in Step 1 in the figure). Instead of CFR, we use tone injection to move along this (blue) PAPR-EVM tradeoff curve. The effect (in Step 2 in the figure) is similar to that of CFR, though: tone injection reduces the PAPR from (only) 9.5 dB to 7.0 dB, but it also introduces some EVM. However, this use of tone injection for a smaller reduction in the PAPR only introduces 2.5% EVM instead of more than 6.0% from the CFR version (from the original 13.0 dB PAPR).

For 6.0%+ EVM (the grey CFR version without SCA above) and a 64-QAM modulation, the figures below demonstrate the EVM at the receiver (mobile device) on the left and the error in the constellation plot on the right.


The left shows the EVM at the receiver, and, more importantly, the right shows the error in the constellation (64-QAM used in this example).

64-QAM cannot accommodate the error above because the data for each point bleeds too much — the receiver cannot discern where a bit is supposed to be in the plot on the right. In practice, the network would use 32-QAM instead of 64-QAM if the EVM were 6.0%+. Compare the images above with the ones below for the signal combined with SCA (2.5% EVM). The plot below on the right shows a cleaner and usable constellation for 64-QAM.


The EVM at the receiver (left) and the error in the constellation (right) for a 64-QAM constellation with 2.5% EVM instead of 6.0% EVM in the previous plots

In this example, the network is able to upgrade the constellation employed for a specific subscriber from 32-QAM to 64-QAM. All subscribers would have a similar upgrade (of one constellation), implying a 2x lift in capacity by adding SCA to an existing CA application of three CCs each of 20 MHz.