Untangling the hot spot backhaul mess

Wireless LAN hot spots face a common issue – how to provide adequate backhaul from each wireless access point onto the public Internet and other WANs. Getting users connected to the wireless network itself is not all that hard for hot spot operators, but carrying the user’s traffic back to the Internet is giving operators all sorts of headaches.

Wireless LAN hot spots face a common issue – how to provide adequate backhaul from each wireless access point onto the public Internet and other WANs. Getting users connected to the wireless network itself is not all that hard for hot spot operators, but carrying the user’s traffic back to the Internet is giving operators all sorts of headaches.

In fact, this backhaul issue is among the most crucial business and technical issues that hot spot operators must overcome to be successful (along with other not-insignificant issues such as security, user authentication and roaming). For today’s hot spot vendors – with the limited usage that many face as pioneers in the field – backhaul is simply too expensive. A coffee shop-based hot spot, for example, may see only a few dozen paying customers a month at $10 a day, or less if these customers subscribe to monthly service plans. The economic issues with this model are plain to see – a $500 per month T-1 service is simply not going to be paid back by this kind of usage, and that’s not even considering any management or equipment expenses.

A ubiquitous hot “zone” deployment in an urban area – with dozens or potentially hundreds of access points – faces this issue on an even greater scale. Providing T-1 or even just DSL connections to all the wireless access points raises the stakes significantly, both in terms of monthly fees and also in terms of simple manageability. And, not insignificantly, the optimal positions for access points in such a network may not neatly overlap the locations reached by the wireline network.

While these logistical and economic issues can be vexing for hot spot vendors today, the future may very well bring with it even greater backhaul issues. First, increased usage will increase the traffic carried over the backhaul network. Even a single 802.11b access point could potentially saturate a T-1 if there was enough user traffic. Secondly, advances in Wi-Fi standards, such as the adoption of 802.11g within the hot spot environment, could greatly increase the backhaul requirements of the hot spot network.

One way to solve this backhaul dilemma is to simply bypass the wireline network for as long as possible by using wireless backhaul to bring hot spot traffic to a smaller number of aggregation points.  An alternative for enabling this wireless backhaul is to use the access points themselves, in a mesh network. An example solution is Tropos Networks, which recently launched its “cellular” Wi-Fi access point product line. Tropos’ access points and Tropos Control Protocol have been designed to autonomously “talk” to each other and provide optimized routing back to those access points that are connected to the wireline network. Although other vendors have offered similar solutions over the past few years, Tropos claims that its control protocol provides “mesh” with a minimal impact on throughput, because the control protocol requires much less overhead than competing solutions.

This kind of meshed network can reduce the backhaul requirements for a hot zone deployment, but may eventually face some scalability issues.  This is due to the fact that the same radio is being used to provide customer-facing bandwidth and to provide backhaul back to the wireline network.  A long-term solution for heavily trafficked networks may be to keep the same basic architecture, but to incorporate a second radio – operating in a different spectrum – for the backhaul traffic.

Point-to-point wireless systems using millimeter wave solutions can provide relatively inexpensive links from access points back to an aggregation point with enough bandwidth to support even the fastest wireless LANs available today. For example, vendors like DragonWave are deploying Ethernet IP-based point-to-point systems with a variety of wireless service providers. Such a system could easily provide backhaul for several access points back to a larger metro aggregation point connected to the wireline network by Gigabit Ethernet or other high speed transport options.

In the long term, perhaps the best solution would be a system that combined the Wi-Fi customer-facing access and the wireless backhaul in a single unit. Such a system would get around today’s backhaul issues and allow a hot “zone” installation to avoid the expense of using the local wireline carrier for individual hot spots.  Aggregation and WAN connectivity could be more centralized and utilize the economies of scale of higher bandwidth Ethernet-based wireline transport solutions.