5G Pushes the Performance Envelope

By Caroline Chan, Wireless access segment manager, Network Platform Group, Intel

Make no mistake about it - there’s a lot of work ahead for telecommunications equipment manufacturers (TEMs) who expect to play in 5G. That’s because in some locations, 5G networks will need to provide “much greater throughput, much lower latency, ultra-high reliability, much higher connectivity density, and higher mobility range,” according to Next Generation Mobile Networks (NGMN) Alliance.

Put another way, “5G is expected to have 1,000 times the capacity of 4G,” said Dr. Xiang Jiying, Chief Scientist at ZTE, in an interview with the Wall Street Journal.

Here are just a few areas where Intel is focusing efforts to help satisfy 5G requirements and the associated challenges.

5G Blog No. 2 image.jpg

Greater base station compute performance

Today’s 4G base stations typically have two to four antenna elements, moving to hundreds with 5G. This will be accompanied by massive multi-user, multiple-input, multiple-output (MU-MIMO) technology that is expected to deliver Gbit per second data rates. These massive arrays will generate ultra-narrow beam patterns, which can be steered towards intended users while simultaneously suppressing energy to unintended users.

“Unfortunately, the promised benefits of large-scale MIMO come at the cost of significantly increased computational complexity in the base station. In particular, data detection in the large-scale MIMO uplink is expected to be among the most critical tasks in terms of complexity and power consumption, as the presence of hundreds of antennas at the base station and a large number of users will increase the computational complexity by orders of magnitude,” writes Michael Wu and others in a paper published in the IEEE Journal.

To address the need for these higher compute levels in a space-constrained form factor, Intel is developing faster processors per Moore’s Law and investigating new instructions that could accelerate MIMO algorithms.

Higher frequency spectrum

The spectrum for 4G is in the range of 700 MHz to 2.5 GHz, but higher frequencies are needed for 5G in order to handle the anticipated traffic increase. “The entire spectrum from 10 GHz up to 100GHz, which is well into the millimeter wave (mmW) range, is being considered for use by 5G mobile communication systems,” suggests Erik Dahlman and others from Ericsson. Since 5G equipment will probably be expected to handle 4G/LTE as well, support for the entire spectrum – from below 1 GHz up to mmW frequency bands –will likely be needed.

Contributing to this effort, Intel along with 15 telecoms operators, vendors, research centers, and academic institutions from eight European countries launched a collaborative project to develop millimeter-wave radio technologies for use in ultra-fast 5G radio networks. Members of MiWaveS, which stands for millimeter-wave small-cell access and backhauling, believe deployment of small cells with millimeter-wave access in dense urban areas will increase the flexibility of the access infrastructure as well as improve spectral and energy efficiency for low-power access points.

At the Mobile World Congress 2015 in Barcelona, Intel showcased a prototype of Anchor Booster, which anchors signaling at a virtualized macro base station and boosts data capacity for small cells that operate at 60 GHz.

Lower edge latency

The user experienced data rate for 5G is expected to take a big jump over typical 4G rates of 5 to 12 Mbps and a 100 Mbps peak rate. According to Ericsson, 5G should be able to provide the following data rates for particular scenarios:

  • Indoor and dense outdoor environments: 10 Gbps and higher
  • Urban and suburban environments: several 100 Mbps
  • Sparsely-populated rural areas (developed and developing countries): at least 10 Mbps

These higher 5G target data rates translate into the need for even lower latency than 4G, with end-to-end latency going down from around 10 milliseconds (ms) to 1 ms or less. one way to address this requirement is to deploy more services at the edge, which can provide users faster access to data. From the perspective of the base station, this means virtualization and interrupt overhead must be cut down by an order of magnitude.

Intel, Wind River, and VMware are working on a series of processor architecture and software enhancements aimed at improving latency and determinism of computing platforms. In addition, Intel co-founded the ETSI Mobile-Edge Computing (MEC) Industry Specification Group (ISG) to help drive technical standardization, and enable IT and cloud-computing capabilities within the radio access network (RAN) in close proximity to mobile subscribers. Since the ISG inception in September 2014, its membership has grown to over 40 companies. Making this easier, the Intel® Network Edge Virtualization SDK (NEV SDK) is an open standards-based solution that delivers a ready-to-use, low-latency, virtualized environment for base stations.

For more information, learn more about Intel’s solutions for wireless infrastructure.