Qualcomm (2016). Making 5G NR a reality. Leading the technology innovations for a unified, more capable 5G air interface. CTIA Super Mobility 2016 – 5G Technical Workshop.
Synopsys Inc (2019). ASIP designer [Online]. Available at: https://www.synopsys.com/dw/ipdir.php?ds=asip-designer [Accessed 4 August 2020].
Werther, O., and Minihold, R. (2013). LTE: System specifications and their impact on RF & base band circuits. Application note, Rohde & Schwarz.
1 For a color version of all figures in this book, see www.iste.co.uk/andrade/multi2.zip.
2 1 According to data (GFK 2019), smartphone and related wearables were a $522 billion dollar market in 2018.
3 2 Loosely translated as a single packet of data, for the broader audience.
4 3 Simplified: TTI consists of OFDM symbols, which, in turn, consist of transformed sets of quadrature amplitude modulation (QAM) symbols, which, in turn, consist of transformed sets of soft bits, which are a mix of guard and information bits.
5 4 TTI is inversely scaled with subcarrier frequency spacing (SCS). In 5G, SCS is base spacing (15 kHz, same as 4G) ×2μ and TTI is base TTI (1 ms, same as 4G) ×2−μ.
6 5 We do not have a reference for this rule of thumb, since the actual timing budget distribution is vendor-specific and kept private. This is a first-order approximation.
7 6 1 kRB = 1,000 RB, 1 RB = 12 subcarriers × 2 slots of 7 (6) symbols per slot = 168 (144) RE, (·) – extended Cyclic Prefix (CP) mode possible for μ = 2 or LTE. An RB is a unit in technical terminology representing a grid section onto which QAM symbols are mapped; one element of that grid is called an resource element (RE).
8 7 Simplified: for a radio link.
9 8 Note that this should not be confused with massive MIMO, which uses constructive and destructive interferences of radio waves from multiple antennas to create a single data stream layer.
10 9 4G modulation code rate schemes (MCS) (3GPP 2019b).
11 10 5G MCS (3GPP 2019f).
12 11 This is the lowest rate per smallest channel. Even though a voice call may require between 8 kb/s and 80 kb/s (depending on the codec (ITU 2019a, b)), the block nature of discrete Fourier transform (DFT) and the required sample rate used in 4G forces the modem to work with higher average rates than those required by the application.
13 12 Simplified: simultaneously active multiple channels on different frequencies.
14 13 An additional motivation for having an extent of SW solutions in communication modems is 1) their ability to be exchanged easily via updates, compared to exchanging HW, and 2) they are ideal for multiplexing a multitude of standards on the same HW.
15 14 A self-contained piece of SW or function that fully captures the functionality of an algorithm.
16 15 Note that the first inner loop will not be executed for l = 0.
17 16 At each stage, quantization occurs anew, for example .
18 17 Accumulator quantization: the result of multiplying Q0.n numbers is Q0.2n, which still represents a value in the range [−1, 1), just with 2n bits, where n = databits. This value is then again requantized with ACCbits = m bits.
2
Towards Tbit/s Wireless Communication Baseband Processing: When Shannon meets Moore
Matthias HERRMANN and Norbert WEHN
Microelectronic Systems Design Research Group, TU Kaiserslautern, Germany
Mobile communication plays a central role in our information society and is a key enabler in our connected world. The newest standard, 5G, features data rates >10 Gbit/s. Beyond 5G, data rates towards 1 Tbit/s are expected. The tremendous improvement in mobile communication has to be considered alongside the progress in the microelectronic industry. For many decades, improvement in silicon process technology provided better performance, lower cost per gate, higher integration density and lower power consumption. However, we have reached a point where Moore’s law is slowing down and microelectronics cannot keep up with the increased requirements coming from communication systems. Thus, the design of communication systems is no longer just a matter of spectral efficiency or bit/frame error rate. When it comes to implementation, design of communication systems requires a cross-layer approach covering information theory, algorithms, parallel hardware architectures and semiconductor technology to achieve excellent communications performance and high implementation efficiency. In this chapter, we focus on channel coding, which is a major source of complexity in digital baseband processing, and will highlight implementation challenges for the most advanced channel coding techniques, i.e. Turbo codes, Low Density Parity Check (LDPC) codes and Polar codes, for throughput requirements towards 1 Tbit/s.
2.1. Introduction
Wireless communication plays a central role in our information society and is a key enabler for our connected world. Today we already have more than 20 billion connected devices, and the majority of them are wirelessly connected. The first generation of mobile communication systems has shifted the communication from landline to handheld devices, followed by the third (3G) and fourth generations (4G) that marked the advent of the mobile internet. We have seen a tremendous increase in data rates over the different generations, for example, the Global System for Mobile Communications (GSM) featured about 10 kbps, the Universal Mobile Telecommunications System (UMTS) about 2 Mbit/s and Long Term Evolution Advanced (LTE-A) about 1 Gbit/s. The newest standard, 5G, features data rates > 10 Gbit/s. Beyond 5G, data rates toward 1 Tbit/s are expected (EPIC 2020). However, increasing data rates are not the only driver. With the advent of 5G, wireless communication is finding its way into a variety of applications with largely diverse requirements. Massive Internet of Things (IoT) demands for extremely low power, massive content applications (e.g. video streaming, cloud office) call for very high throughput and the use of wireless communication in control applications (e.g. autonomous driving, remote control, tactile internet) requires very fast response times in the order of 1 ms. Thus, communication has to support a huge throughput–latency range (5 orders of magnitude in throughput, 3 orders of magnitude in latency) and a large diversity of applications with different Bit Error Rate (BER)/Frame Error Rate (FER) requirements (Fettweis and Matus 2017). In this chapter, we focus on