Large macro base stations companies have a seemingly attractive offer for wireless operators – “Just use our macro base stations indoors! We can offer you new indoor radio heads that connect to the baseband/digital units you love, and you will be on your merry way.” The problem with this offer is that it is trying to force-fit technology designed for towers and other outdoor environments – called CPRI – into an enterprise environment.
CPRI is a standard designed to carry baseband I/Q samples from the baseband unit to a radio head. CPRI was developed so that base station OEMs could source radio transceiver units from multiple suppliers. Over time, CPRI made it possible for the radio transceivers (radio heads) to be moved from the base station chassis to the tower top, with a fiber optic cable in between. This move to remote radio heads reduced cable losses, and boosted the coverage of macro base stations. All good things.
However, CPRI was never designed to be bandwidth efficient. A single 20 MHz LTE carrier with 2×2 MIMO requires 4.9152 Gbps of bandwidth to carry CPRI between the baseband unit and each radio head*. Considering that a 20 MHz LTE carrier can deliver up to 150 Mbps on the downlink and 50 MHz on the uplink, CPRI’s bandwidth efficiency is less than 3%. In practice, it is even lower. Some vendors compress CPRI, but compression adds cost and barely increases CPRI efficiency by a factor of 3.
The CPRI approach works fine when there is dedicated fiber between the base station and each radio head, but is not appropriate for buildings. Networks inside buildings are built with switched Ethernet. Enterprises want to connect all kinds of equipment to their Ethernet LAN – from computers and printers to Wi-Fi APs and small cells. The Ethernet LAN is the “neutral host” network infrastructure for the enterprise, and all these devices need to share the bandwidth on the LAN. Over the last 20 years, a wide range of technologies have been developed to share the LAN, from QoS to VLAN, and they were developed for a good reason: it does not make sense for enterprises to go around deploying custom cable infrastructure for every new technology they need.
Some vendors are proposing non-CPRI approaches that split baseband processing between the baseband unit and the radio head, and are able to work on Ethernet instead of fiber. Such approaches require slightly less than 1 Gbps per radio head to carry a 20 MHz channel. Still, the throughput required on LAN increases linearly with the number of radio heads connected to the baseband unit, making it unfeasible to share the enterprise LAN.
SpiderCloud’s scalable small cell system is the only system that can serve thousands of subscribers in buildings as large as 1.5 million square feet and works on a shared enterprise LAN. In a SpiderCloud system, all baseband processing is done by its small cells, called radio nodes. Up to hundred dual carrier radio nodes are connected to SpiderCloud’s small cell controller, Services Node. The Services Node is responsible for anchoring all the user sessions, managing mobility and interference, and SON. This architecture ensures that the only traffic between SpiderCloud’s small cell and Services Node is actual user traffic, and some few hundred kbps of overhead. In other words, deploying a SpiderCloud small cell system on an Enterprise LAN is no different than deploying an enterprise Wi-Fi system. SpiderCloud’s ability to share the enterprise LAN is one of the main reasons why Cisco selected SpiderCloud to build 3G and LTE modules that clip on to Cisco’s market-leading Aironet Wi-Fi access points.
Finally, let’s consider what happens in the future with LTE-Advanced and 5G. With LTE-Advanced and 5G, operators are looking for ways to use more spectrum and more antennas. The amount of bandwidth required for CPRI scales linearly with channel bandwidth, number of channels and number of antennas. If CPRI-based approaches are struggling to, but are unable to offer 20 MHz LTE on Enterprise LANs, how will they offer LTE-Advanced and 5G? Perhaps, this is the reason why NGMN and many other organizations looking at the future of mobile networks, believe that densification using small cells is the way forward.
* This is how the math works: 30.72Msps (sampling rate) * 15b (sampling width) * 2 (I/Q) * 16/15 (1 control word per 15 data words) * 10/8 (CPRI line coding of 10b/8b) * [2 (for 2 Tx antennas) + 2 (for 2 Rx antennas)] = 4.9152 Gbps. Assuming peak cell throughput of 150Mbps in DL and 50Mbps in UL, the best-case efficiency of CPRI (bandwidth on CPRI link / actual throughput delivered by cell) is less than 3%!
– Amit Jain, Vice President of Marketing & Product Management