It seems the race to demonstrate raw bit speed under the early 5G new radio (NR) is starting to heat up. Huawei presented a scenario that claimed better than 32 Gbps download speed at the recent PT/Expo in China. But what does that exactly mean?
Raw speed has its place in certain areas like motor sports. They who go the fastest get the trophy, simple as that! However, port that to the electronics segment and the lines get fuzzy. For example, raw CPU speed isn’t necessarily the determining factor in performance when it comes to computers.
In fact, raw CPU cycles are not even up there in bragging rights anymore. With computers, throughput is usually measured in task performance. Running a CPU benchmark and getting a number is meaningless if the software has sloppy code and the rest of the hardware, peripherals, memory, core I/O mapping, the base hardware platform, etc., are not top shelf. The same parallel can be drawn in wireless systems.
So to say Huawei is now the leader, because they managed to get a 32 Gbps peak download speed, doesn’t really tell the story. It is just good for publicity to keep the 5G train moving.
As with most of this type of data from China, the details are few. All Huawei says is that “Using 200 MHz bandwidth on the C-band, the downlink cell peak rate exceeded 32Gbps. Huawei assumed the initiative to integrate all key 5G New Radio (NR) technologies. These include flash orthogonal frequency-division multiplexing (f-OFDM), new frame structure, new codes (such as Polar Code), new parameter sets, massive multiple-in, multiple-out (MIMO), and sparse code multiple access (SCMA).” OK, but how did they do all of this? I could not find any technical details about any of this.
Since this is set up under the “non-standard” spec., it would be nice to see the parameters of the system and what part of the non-standard stuff was actually used (stuff is occasionally a technical term, btw).
In cases so far, such test results were obtained under optimal conditions with tweaked equipment pushing the parameter limits – something not likely to be implemented in GA products. They also claimed 300 ns latency (one-way), but was that consistently or sporadically?
It is no secret that the Chinese like to push the limits of the edges and spin it to make it sound like it is the greatest advance in the history of electronics. I am not judging, just saying what I have experienced. Raw numbers like this mean little to me.
There is little doubt that the faster the download speed, or the wider the bandwidth, the more data can be packed into the pipe. Therefore, having a fat pipe is, without a doubt, the goal. But with that has to come a lot more than just speed. All the speed in the world isn’t going to matter if the traffic and bandwidth is not optimized.