One of the largest hurdles to deploy outdoor small cells is the total cost of ownership of the entire solution of which backhaul comprises a large part. Hence, cost-effective backhaul is one of the main prerequisites for small cell deployments. While most wireless backhaul solutions today are point-to-point (PTP), the use of intelligent point-to-multipoint (PMP) wireless backhaul becomes a necessary step in the evolution of wireless networks from a macro-cellular architecture to one of heterogeneous network comprising multiple layers of base stations even ones with different radio access technologies.  


Multi-point wireless backhaul systems, as shown in Figure 1, affect the operator business case on both the capital and operational expenditure sides. From a capex perspective, it involves a reduction in number equipment as a single backhaul hub module connects to multiple remote modules, each of which is collocated with a compact base station. Larger savings are realized as the cost of backhaul of data aggregated at the hub back to the core network is reduced with the greater number of remote multi-points.


What makes PMP backhaul a possibility in small cells is the fact that small cell traffic is very ‘lumpy.’ Because of their small coverage area, small cells tend to have few users most of the time, with peaks at certain times during the day due to crowds. This in turn results in typically low traffic with very large spikes during specific hours of the day when the area is crowded, as shown in Figure 2. The result is high peak-to-average traffic ratio on small cells. In comparison, macrocells have low peak-to-average traffic ratio as in Figure 3 and Figure 4.





Isolated small cells tend to display larger peak-to-average traffic ratio than small cells that are situated in open areas. In the latter case, interference works to limit the maximum achievable capacity. Hence, a user experiences the highest throughput performance when there are few users on the same serving cell as well as in adjacent small cells. In either case of lightly loaded or heavily loaded cells, user traffic itself is bursty due to the nature of data which is packetized.  

Data traffic characteristics means that multipoint backhaul is architecturally viable for small cell backhaul. When few users are present and it’s more likely for a remote node to peak, it is statistically unlikely that all remote nodes will peak at the same time. When the area is crowded, interference will limit the throughput to a mean value that is a few multiples below that of the peak throughput allowing in the process the backhaul of several remote base stations simultaneously.

To improve on the performance of multipoint backhaul, an additional element is implemented: intelligent traffic management. Intelligent traffic management has two main components: the first is dynamic bandwidth allocation; and the second is Quality of Service (QoS) management.   

Dynamic bandwidth allocation implies shifting of backhaul hub resources from one multipoint to another depending on traffic requirements.  The scheduler function at the hub plays a critical part in tis since traffic resource allocation is time sensitive: capacity needs to be provided when it is needed; otherwise, it will result in lost data packets. Furthermore, traffic requirements of one multipoint needs to be weighed to that of another. This is where QoS management comes in.  

To better understand QoS management, we note that backhaul data consists of user data plus overhead which comprises signaling & control communication, management communication and synchronization. It is estimated that control plan signaling is up to about 20 percent of data plane traffic although often times higher ratios are planned (especially when IPSec is used). Each type of traffic has unique characteristics that are defined by certain parameters such as latency and jitter. Intelligent traffic management entails the ability to detect and prioritize traffic according to its classification parameters. For example, VoIP traffic which is part of the user data is very sensitive to delay but requires low bandwidth. Web browsing and email traffic is more tolerant to delay and jitter but require higher throughput capabilities. Similarly synchronization signaling has low delay tolerance as do many other forms of control and signaling traffic. For this reasons, the ability to support multiple services each with defined parameters becomes an important part of the backhaul solution, particularly when it is coupled with dynamic bandwidth sharing. In this case, the backhaul hub can reserve minimum capacity for each multipoint which is used for delay sensitive traffic while delay tolerant traffic can be prioritized on a best effort basis.

To conclude, scaling the capacity of next generation networks through small cells requires innovative and flexible backhaul technologies that minimize the total cost of ownership to the operator. Intelligent multipoint backhaul combines PMP technologies with intelligent traffic management to minimize the cost of backhaul while maintaining the necessary performance expected by the network operators. 

 


Edited by Stefanie Mosca