There is much debate in the industry about how to converge the voice traffic from existing cellular base stations with the high growth data traffic that is being driven by the rapid adoption of the mobile internet. The volume of traffic, even on 3G networks, is being dominated by data; however the base station interfaces by and large remain TDM based (either E1 or T1 running ATM IMA traffic). This problem is compounded by the introduction of new Radio Access technologies such as LTE (News - Alert)and WiMAX that are intrinsically IP based, which has lead to a debate in the industry about how to backhaul the IP and E1/T1 traffic. The bulk of the debate has been centered around hybrid systems that carry the TDM and IP traffic in native format in parallel across the network, and IP based systems that convert the TDM traffic to packets at the edge of the network in order to transport all traffic across a single IP based backhaul. Although there are pros and cons of each of these approaches, neither of them actually delivers the most efficient network.
Mapping 3G base station traffic, whether it is voice or data, into ATM cells, then into E1 interfaces and finally into SDH results in 30% of the network capacity being wasted in overhead and protocol conversion. This is magnified by the fact that many of the E1s are only partially filled since the total interface capacity must be sized for the peak load, not the average load. Secondly, different types of traffic are often segregated into separate E1 interfaces - resulting in further stranded capacity due to partially filled E1s. The inefficiency due to partially filled interfaces can easily be an additional 50% on top of the protocol conversion inefficiencies. As a consequence, operators must often provision 3 times more backhaul capacity than is actually used on average. This is true whether one is using either hybrid TDM + IP systems OR pseudo wire + IP backhaul solutions.
It would seem obvious that converting the voice and data to packets inside the base station would result in a much more efficient network. This approach would eliminate the protocol conversion inefficiencies as well as the inefficiencies due to partially filled E1s. Statistical multiplexing can then be employed to mitigate the peak traffic loads, further reducing the total backhaul capacity required. Finally, the backhaul becomes much simpler and cheaper - a flat IP network end to end without the additional cost of a parallel TDM network, or the cost of pseudowire devices.
Convergence (News - Alert)inside the base station is resonating with the service providers and, according to Rethink Research, two-thirds will now only buy Ethernet capable base stations for new deployments and 45% aim to phase out non-Ethernet base stations from 3G/4G networks within four years. In 2009, there were 900,000 new cell site connections, which were mainly Ethernet enabled or upgradeable.
The issue that is most often cited as a concern for going to an all IP backhaul network is synchronization. Advances in standardization of packet based techniques (1588 v2) and line timing based techniques (synchronous Ethernet) coupled with real world success stories (for example WIND mobile in Canada who have deployed an all IP 3G network with 1588 v2 synchronization distribution) have largely laid that concern to rest. As there are more success stories, more operators will feel comfortable with the new technology and we can expect to see a rapid adoption, especially for HSPA+ and LTE deployments.
For the cases where 2G base stations are co-located the one or 2 E1s required to service the 2G traffic can be handled in a variety of ways. Some base stations, like those from Nokia Siemens (News - Alert)Networks, can provide Ethernet interfaces even off of the 2G base station. Other carriers have adopted a strategy of deploying cell site routers in order to have a uniform provisioning and demarcation platform across their network. Most cell site routers also support pseudowire interfaces and can convert the E1s to packets. Finally, in regions such as North America, where leased copper circuits are the predominant historical backhaul mechanism, these can be left in place for the 2G base stations.
The hybrid vs pseudowire debate does seem to miss the point. The obvious efficiency, simplicity and economic advantages of using a flat IP transport network connected to packet interfaces directly off the base station seems to be clear. The trigger for this shift will be confidence in the synchronization coupled with the deployment of new technologies such as HSPA+ or LTE, or the roll out of new applications which drive incremental capacity on the network such as the recently announced unlimited Netflix on the AT&T (News - Alert)network, or the 3G-enabled iPad. Since the majority of the new base station connections simply require a software change to enable the Ethernet ports on the base stations, the shift to packet interfaces directly off the base station may happen much more quickly than any of us are currently thinking.
TMCnet publishes expert commentary on various telecommunications, IT, call center, CRM and other technology-related topics. Are you an expert in one of these fields, and interested in having your perspective published on a site that gets several million unique visitors each month? Get in touch.
Edited by Patrick Barnard