5G/vRAN Basics Explanation by Dmitry Lachover

There is a significant trend in the RAN (Radio Access Network) market going from traditional vertically integrated RAN equipment towards a more open vRAN (virtualized RAN) and ORAN (Open RAN). The vRAN/ORAN efforts bring in some new usage models and technologies that did not exist in traditional RAN, e.g.  open standards based definitions of RU, DU and CU, Containerized deployments for DU and CU, Kubernetes based orchestration, Cloud, etc

The objective of this article is to give a brief into the basics of vRAN for people who are new to this domain.

Although participants in 5G projects acquire knowledge about cellular communication from different sources many questions still remain

This article attempts to answer frequently asked questions at deep level

• 4G versus 5G

The RAN (4G) is the final link between the network and the phone. It includes the antennae we see on towers and on top of buildings connected to Basestation Units (BU).  When we make a call or connect to a remote server e.g. to watch a YouTube video, the antenna transmits and receives signals to and from our phones or other hand-held devices. The signal is then digitalized in the RAN BU and connected into the network.

Operators today does not want proprietary functionality of the equipment-makers, operators want  more diverse ecosystem of vendors like:

1. Fully decoupled software running on abstracted general-purpose hardware
2. Functional components implemented as abstracted SW interacting via standardized interfaces 
3. Cloud-native capabilities in deployment

In an Open RAN environment, the RAN is disaggregated into three main building blocks: Radio Unit (RU), Distributed Unit (DU), Centralized Unit (CU)

The RU is where the radio frequency signals are transmitted, received, amplified and digitized. The RU is integrated into the antennae. The DU and CU are the computation parts of the base station, sending the digitized radio signal into the network. The DU is physically located at or near the RU whereas the CU can be located nearer the Core Network.

The key concept of Open RAN is opening the protocols and interfaces between these various building blocks (radios, hardware and software) in the RAN:

·       Fronthaul between the Radio Unit and the Distributed Unit

·       Midhaul between the Distributed Unit and the Centralized Unit

·       Backhaul connecting the RAN to the Core Network

 

The described change is shown at the picture below – Old 4G vs. New 5G concept

5G new functional splits and Interfaces

 

During a call at 5G Uplink, data is transmitted over “air” from mobile phone (called UE at 3GPP) to Radio Unit (Antennae) then delivered via fronthaul by means of eCPRI protocol (RFoE – RF over Ethernet) to DU – Distributed Unit which performs most of the baseband processing (big part of L1 PHY processing, MAC/L2 and RLC/L3 processing) and then DU delivers the data for further processing to CU – Central Unit via Midhaul

In Downlink, data propagates just in the opposite direction

DU is actually main module that constitutes Basestation in terms of 4G where DUs are actually cloud for RUs

Deployment of RUs, DUs and CUs constitutes vRAN (virtual Radio Access Network) –                  RAN Disaggregation and Distributed deployment

Having it in mind we can talk about meaning of basics

• Layer, mMIMO and Beamforming

MIMO (multiple-input and multiple-output) is a method for multiplying the capacity of a radio link using multiple transmission and receiving antennas to exploit multipath radio signal propagation

Let’s take first a regular 4x4 MIMO: 

Each Tx antenna is actually a layer that transmits data stream, thus 4 Tx antennas transmit 4 different data streams over 4 different “layers”. In other words every stream of data sent via a transmitting antenna in regular 4G MIMO (or via a beam in 5G) is referred to as a “LAYER”

Of cause there is an option to transmit the same data over all 4 layers/Tx antennas, thus the transmitted data rate may be raised dramatically since it has 4x more transmission power for the same data and the much better SNR – this is the tradeoff

However typically different data is transmitted over different layers

mMIMO - massive MIMO uses a lot of TX antennas (hundreds) when it focuses the energy radiated over a transmit antenna array (a subset of antennas) in a certain direction

So in mMIMO each beam is actually a layer, meaning the number of different data streams (layers) equal to the number of beams while each beam created by a number of antennas (by means of special complex data precoding at each antenna that “creates direction”)

Thus for example “64T64R (16DL, 8UL layers)” means 64 antenna Transceiver, however what important in terms of amount of transmitted data is not a number of Tx antennas but number of layers

In the example above, we are talking about 16 DL layers (16 data streams from DU/RU to UE) and 8 UL layers (8 data streams from UE to RU/DU) where DU is Distributed Unit and RU is Radio Unit

In short, 4G Basestation Unit (BU) is split to DU, CU (centralized unit) and RU along with split of BU PHY processing between RU, DU and CU (O-RAN split 7-2 option for example)

• Carrier Bandwidth

In the example below the carrier is Fc (let’s say 2Ghz) and Carrier Bandwidth is 2*FBW (it may be  400Mhz or 200Mhz or 100Mhz or 20Mhz in 5G)

The carrier bandwidth is divided by subcarriers (in 5G each subcarrier maybe 15Khz or 30KHz or 60 Khz)

Each such subcarrier is the basic instance that caries data ( N QAM modulation defines how many bits are transmitted over each subcarrier)

There is some control procedure of “understanding” the quality of channel/transmission and then MAC (L2) layer decides for example to reduce modulation order, and delivers the new modulation parameters to PHY Layer via FAPI L2-L1 interface. 

The higher modulation the smaller distance between symbols and then smaller noise will easier convert some symbol to the neighbor symbol

Now, a subset/chunk of these subcarriers are assigned to each user, while parts of this chunk are allocated to each channel of the given user

Then, a data over subcarriers is transmitted over time where the basic time unit is OFDM symbol meaning that at the “1^st symbol time” some data is transmitted over each subcarrier, at the “2^nd symbol time” other data is transmitted over the same subcarriers and so on.

You can see at the picture above:

At each symbol, data is transmitted on all frequency subcarriers

Let’s say you have 8 subcarriers, each subcarrier carries 6 bits, so during  4 symbols, 192 = 8*6*4 bits will be transmitted where time duration of each symbol is ~70us (14 symbols in 1ms) , meaning 8*6*4 bits will be transmitted during 4*70us=280 us

• Subframe

There is standard 4G/5G “frame”, “subframe” and “slot” definition each for its processing purpose

In the current context, what important is 1 subframe (1ms) = 14 Symbols. 1 subframe consists of 2 slots where each slot consists of 7 symbols.

One Transport (data) block is transmitted during 1 subframe where 14 symbols are used to transmit this transport block. Next 14 symbols will be used to transmit next transport block

• Subcarriers assignment 

A subset/chunk of subcarriers assigned to each user is divided in its turn for data and control channels in frequency and time

At the picture below, the orthogonal axis is part (let’s say 1Mhz) of the total Carrier frequency (let’s say 100Mhz) that is given to 1 user

(for example , 1 Mhz out of 100Mhz is given to 1 user)

If one subcarrier is 15Khz wide then N=1Mhz/15Khz subcarriers are assigned for this user

You can see that 72 subcarriers are assigned for yellow (PRACH) channel and this channel is transmitted all over all the symbols (all the time)

Different channels are used for different purposes, most of the are kind of control channels

For example RS (reference signal) used for equalization purposes takes some number of subcarriers (let’s say 72*4=288 subcarriers) and transmitted over symbol number 3 and number 10 in time

BTW, for example, RS channel’s data does not propagate outside the PHY layer (RS is purposed for PHY processing only)

What important in traffic context is “light blue” data channel (PUSCH) that takes let’s say 288 subcarriers for each user and transmitted over 12 symbols in time during each subframe (subframe = 1ms = 14 symbols)

So all users together will be transmitted over 12 symbols in time and actually will use most of the total number of subcarriers (most of the total carrier bandwidth)

• 1 Cell is actually 1 carrier bandwidth space and is actually 1 RU

At this picture BU may be replaced by DU

In 5G, DUs are actually Virtual Functions (VFs) managed by Kubernetes for example on some DMS (deployment management service) platform. RU "talks" to DUs cloud via fronthaul network

• TDD - Time Division Duplex, FDD - Frequency Division Modulation

In TDD mode, both uplink and downlink use the same spectrum frequencies but at different time

Typically DL takes 4/5 of the total time and UL 1/5 of the time 

In FDD mode, both uplink and downlink are transmitted at the same time (all the time) at different spectrum frequencies

Understanding the meaning of each parameter we can do data traffic calculation between RU and PHY Layer (L1) of DU

Amount of data per 1ms = Number of subcarriers * Number of Modulation bits per subcarrier * number of data symbols in subframe (1ms) *  Number of Layers * Number of Cells

Number of Modulation bits per subcarrier: 1024 QAM = 10 bits, 256 QAM = 8 bits, 128 QAM = 7 bits, 64Q AM = 6 bits

• L1 - L2 (MAC/RLC - PHY) Interface 

L1 and L2 (MAC/RLC and PHY) layers inside DU communicates between them by means of FAPI (Femto API) interface. The amount of data between L1 and L2 is reduced by a factor of coding rate

Coding rate means how many parity(extra) bits are added per 1 bit of information. For example 1/3 coding rate means 3 data transmitted bits to air per 1 actual info/raw bit. Typical coding rates are 1/3, ½, 2/3, ¾, 4/5, 5/6 (higher modulation, more noisy channel and then stronger coding rate is required)