Hosted Communications

Historically the interfaces used on CDMA and GSM base stations have conformed to Time Division Multiplexing (TDM) standards, either T1(1.544 Mbps) or E1 (2.048 Mbps). As a result of this, the backhaul for mobile networks has been dominated by PDH/SDH microwave or leased line solutions, which were designed around the transport of these types of interfaces. Although-IP based backhaul for the current Radio Access Network (RAN) has had some success simply due to the lower cost of Ethernet transport solutions, the majority of the network still utilizes circuit switched technology. There are three compelling events, however, that are driving the evolution to Ethernet/IP backhaul networks.

These events will accelerate the adoption of IP backhaul for the RAN, making it not a question of if but of when a carrier must adopt this technology. The question facing every carrier then is when they spend the next dollar on infrastructure are they going to spend it optimizing the delivery of legacy services or start optimizing their network for the coming services?

For many carriers the adoption of IP backhaul represents not only the usual technology risk, but also a need to train and educate the network designers and operations staff. The very elements that make IP networks attractive — statistical multiplexing of bursty traffic, connectionless networking, and variable bandwidth require a new approach to designing and operating the networks. Couple with that the need to still preserve the deterministic, high-quality circuit transport for the legacy services that are paying today’s bills gives many operators headaches, especially those that have not operated or owned their own backhaul infrastructure before.

Given this level of uncertainty, some operators may perceive a hybrid approach where the TDM and IP are carried over the same infrastructure but in their native formats as lower risk. What this approach ignores, however is that by adopting this seemingly lower risk approach the carrier is in fact forgoing almost all of the advantages of a converged network. Even though a single wireless link may be able to transport both TDM and Ethernet, the operator still must deal with both traffic types at each intermediate node. This drives the need to place separate aggregation, grooming, redundancy and management solutions for each traffic type in the network, increasing both the capital cost as well as significantly complicating (increasing the cost) operation of the network.

An alternative approach to this problem is to deploy an IP backbone and utilize pseudowire technologies to packetize the TDM traffic at the edge of the network. This does deliver on the benefits of convergence. At each intermediate node in the network all the pass-thru traffic remains packet based, dramatically simplifying the interconnect cabling and reducing the need for TDM or ATM platforms distributed throughout the network. This pass-thru economic benefit, coupled with the increased capacity available to IP networks (they are not limited to integral multiples of the SONET/SDH data rates, but can take full advantage of the RF channel bandwidth) means that pseudowire networks are in fact more efficient and lower cost than their equivalent hybrid networks.

If the pseudowire network architecture is so compelling, then why is this not widely deployed? The principal reason is that the drivers that are forcing carriers to move to IP networks are just beginning to be felt by the broader market. Like the early days of VoIP, some carriers are wary of adopting the new technologies until they are more widely deployed. As with VoIP, it will take some time before everyone is convinced — both of the economic benefits of the pseudowire architecture and the viability for demanding backhaul applications. And just like VoIP, in the years to come it will be very few people who can even recall why they used to do things the old way.