Maravedis: Filling fiber up with baseband backhaul

By Esteban Monturus, Market Analyst - Europe & Backhaul, MaravedisAs we at Maravedis have long warned, average access speeds for LTE subscribers under normal network load conditions at random locations within the coverage area are considerably lower than the 100 Mbps downlink and 50 Mbps announced by some operators. Last week, PC World magazine published some results after having tested the most advanced mobile broadband technology from all the tier 1 operators across 260 locations in 13 US cities. The results show Verizon's LTE as the winner over T-Mobile's and AT&T HSPA+ and Sprint's WiMAX, averaging 6.44 Mbps downlink and 5 Mbps uplink.

Considering the fever to get fiber to the LTE base station by some early LTE adopters, what are those tens of Gbps needed for? Today providing fiber backhaul to LTE sites is more related to a willingness to concentrate efforts on the RAN segment, discarding any backhaul issue at the expense of not fully using resources. However, in the long term operators should take additional advantage of this backhaul capacity. In my opinion, innovations from vendors such as Alcatel-Lucent, with its LightRadio concept, and from operators such as China Mobile with its C-RAN architecture, are providing an answer. Both LightRadio and C-RAN enable virtualized RAN: concentrating network baseband processing power from many cell sites into a few locations in order to lighten physical hardware needs at the point of radio delivery. The higher computing efficiency of such central clusters would considerably reduce RAN power consumption, representing a high percentage of any operator's power consumption (73 percent in the case of China Mobile).

Virtualized RAN architecture would enable for additional OPEX savings in operations and maintenance (traveling expenses are reduced), and in site rental (only all-outdoor radio equipment needs to be installed). CAPEX would be reduced as well due to lower site acquisition and civil works costs. China Mobile thus estimates virtualized RAN architecture would enable a reduction of up to 50 percent in OPEX and 15 percent in CAPEX in typical dense urban scenarios. Furthermore, virtualized RAN would make advanced techniques such as Coordinated Multi-Point transmission/reception (CoMP) and dynamic inter-cell resource sharing easier to implement.

In terms of the backhaul capacity requirement, shifting baseband processing from the cell site to a unified location implies IP backhaul to become "baseband backhaul." That is, IP packets are not longer being transmitted to the cell site, but complex baseband samples which technology agnostic radio units would turn into the analog signal being finally exchanged through the air interface. It is important to note that unlike IP backhaul, baseband backhaul scales with the number of antennas used at the cell site. In addition, because of the so-called Nyquist-Shannon sampling theorem, perfect reconstruction of an analog signal is possible provided that the sampling frequency used is greater than twice the maximum frequency of the signal being sampled. Skipping complicated calculations, a 20 MHz 4x2 MIMO FDD-LTE 3-sector site would end requiring about 16.5 Gbps backhaul capacity, which is much better suited to multi-Gpbs fiber capacity than IP backhaul requirements (which are in the order of a few hundreds of Mbps in the most demanding case).

Given that wireless backhaul technologies allow for capacities in the range of 400 Mbps to a few Gbps, what would be their future role in a heavily virtualized RAN network? I believe wireless backhaul technologies would remain crucial to connect cells deployed in so many locations that would still lack fiber connectivity. The lower backhaul capacity available would not allow those cells to be virtualized, but thanks to all-outdoor compact design they would lower costs compared to most of today's macro cells, though keeping similar performance in terms of coverage and capacity. I do not foresee heavily virtualized RAN before 2020, but it is good to know there is a way to align technological capacity and applications' requirements also for fiber deployments.

Esteban Monturus is a market analyst for Europe and backhaul for Maravedis. Maravedis is a leading analyst firm focusing on disruptive technologies including smart networks using 4G, LTE and WiMAX.

Suggested Articles

With CBRS entering the commercial phase, entities like Google are emphasizing products for the entire ecosystem, not just the SAS part.

Thomas Marzetta, originator of Massive MIMO technology, was appointed director of the research center at the NYU Tandon School of Engineering, replacing Ted…

T-Mobile executives reiterated their plan to use the proposed merger with Sprint to leapfrog competitors AT&T and Verizon in 5G.