A base station has to move the signal between the baseband processing and the antenna, and the antenna usually sits at the top of a tower. The RRH, or Remote Radio Head, changes where that long run of cable goes and what it carries. I'll start with what the RRH does and what is inside it, then explain why operators use it, and finish with what the optical link between the RRH and the base station has to carry.
What does an RRH do ?
Before we look at the reasons for an RRH, we need to know which functions it takes over from the base station. The short list below names them, and the drawing after it shows where each one sits inside the unit.
RRH stands for Remote Radio Head. It is a special component sitting on top of cell tower that mainly performs following procedures.
i) Convert optical signal to electrical signal
ii) Convert electrical signal to optical signal
iii) Up/Down conversion
iv) Amplification of RF signal
The most important feature of RRH is the item i) and ii) meaning that eNB/BaseStation can send and receive data directly to the top of the tower via optical fiber which has almost negligible energy loss comparing to the conventional method that use thick RF coax cable.
The drawing below shows the same tower at two scales. On the right, three radio units sit on the tower top next to their antennas, and two of them carry the label RRH. The optical fibers from the three units join and run down the tower to the Base Transciever Station at the bottom. The fiber label says that the fiber carries baseband IQ. On the left, the inside of one RRH is opened up into its blocks.
Figure 1. RRH on the tower top. The fiber carries digital baseband IQ, so the RF signal exists only in the short path between the RRH and the antenna.
- The Optical Module blocks on the left edge are items i) and ii). They turn the light on the fiber into an electrical bit stream, and the other way round.
- The Baseband to RF Converter block holds the ADC, the DAC and the PLL. Together with the LO below it, this block does item iii), the up and down conversion between baseband IQ and the carrier frequency.
- The PA amplifies the transmit signal on its way to the antenna. This is item iv).
- The LNA amplifies the weak receive signal that comes back from the antenna, before it reaches the converter.
- At the bottom, a second optical fiber leaves the Base Transciever Station. It connects the base station to the rest of the network and is a different link from the one that runs up the tower.
Let's compare this with the list of four items above. Items i) and ii) are about the fiber, and items iii) and iv) are about RF. So an RRH is the complete RF part of the base station, moved up the tower, with a digital fiber interface in front of it. The baseband processing, such as the OFDM modulation and the channel coding, stays in the unit at the bottom.
An RRH is the RF part of the base station, placed next to the antenna : it holds the converters, the LO, the PA and the LNA.The fiber carries digital baseband IQ : it does not carry an RF signal.The RF path becomes short : it runs only from the RRH to the antenna next to it.The baseband processing stays in the unit at the bottom of the tower : the RRH does not decode or schedule anything.
Why we use RRH ?
As I briefly mentioned above, the biggest motivation is to reduce the energy loss while transfering the signal back and forth between the antenna (top of the tower) and processing unit at the bottom of the tower.
In conventional base station, they usually put Amplifier at the bottom and transfer the signal via thick Coax cable. But the pathloss of the coax cable is relatively large, they had to amplify a lot more to compensate energy loss by Coax cable.
To reduce this problem, an idea of putting Amplifer unit at the top of tower (this unit is called TMA(Tower Mounted Amplifier)), but I don't think this idea were adopted widely.
Then RRH was introduced and I think it has been so successful.
The loss hurts the two directions in different ways, so let's look at them separately. In the downlink, every dB lost in the cable is a dB of PA output that never reaches the antenna. The base station has to make up for it with a larger PA, and a larger PA uses more power and produces more heat. In the uplink, the effect is larger. A cable in front of the first amplifier adds its loss directly to the noise figure of the receiver. So a weak UE signal becomes harder to detect, and the uplink coverage shrinks.
This is also the reason behind the TMA. A TMA puts an LNA at the top of the tower, so the receive signal is amplified before it enters the long cable. The RRH solves both directions at once. The PA and the LNA are both next to the antenna, and the long run between the tower top and the bottom becomes a fiber that carries bits. Loss on that fiber does not change the RF signal level at all. It only has to stay small enough for the optical receiver to recover the bits.
Coax loss in the downlink wastes PA power : the base station needs a larger PA to put the same power into the antenna.Coax loss in the uplink raises the noise figure : loss in front of the first amplifier reduces the uplink sensitivity.A TMA fixes the uplink only : it moves the LNA to the tower top, but the PA stays at the bottom.An RRH fixes both directions : the PA and the LNA sit next to the antenna, and the long link is digital.
What does the optical link have to carry ?
Moving the RF part up the tower has a cost. The fiber now has to carry the baseband IQ samples of every antenna, and that stream is much larger than the user data inside it. Before you plan the fiber, you need an estimate of that bit rate.
Let's work out one example for a 20 MHz LTE carrier. 36.211 expresses time in units of Ts, and a 1 ms subframe is 30720 Ts. So the natural sample rate for 20 MHz is 30.72 Msps, which matches a 2048 point FFT with a 15 kHz subcarrier spacing. Each sample has an I part and a Q part. Now assume that each part uses 15 bits. Then one antenna needs 30.72 Msps x 2 x 15 bits, which is 921.6 Mbit/s. The link also carries control words and line coding, so the real line rate is higher still.
Two points follow from this calculation. First, the rate grows with the bandwidth and with the number of antennas. Two antennas double it, and more antennas multiply it again. Second, the rate does not depend on the traffic. The RRH needs a sample every Ts even when the cell carries no user data, because the IQ stream describes the waveform and not the data inside it. This is why the link between the RRH and the base station is always fiber, and why its capacity is planned per antenna and per carrier.
A 20 MHz LTE carrier uses 30.72 Msps : this follows from Ts in 36.211, where one subframe is 30720 Ts.One antenna needs close to 1 Gbit/s of IQ with 15 bit samples : 30.72 Msps x 2 x 15 bits is 921.6 Mbit/s, before control and line coding.The IQ rate scales with bandwidth and antenna count : it does not scale with the traffic in the cell.An idle cell loads the fiber as much as a busy cell : the RRH needs the full sample stream at all times.
Reference
- 36.211 v19.3.0 : E-UTRA Physical channels and modulation. Clause 4 Frame structure, for Ts and the 30720 Ts subframe.