Syncrhonized PulsAR Networks (SPAN)

Jam resistant wireless networks in the unlicensed band

Introduction

With the proliferation of unlicensed radio devices it can be a challenge to deploy and operate dependable RF networks in the unlicensed ISM bands.  The major challenge is, of course, interference from other devices operating in the same band.

What is usually overlooked is that, as your network grows, the sources of interference include your own radios as well as third party devices in the region.

The Afar pulsAR Wireless Ethernet Bridges were designed from the ground up to combat interference. These radios are the building blocks to deploy wireless networks in a multitude of topologies including Point-to-Point, Point-to-Multipoint, Mesh/Tree and Linear Network.  These topologies can be mixed to form very complex and growing networks.  As your own network grows the chances of your devices becoming the sources of interference also grows.

Our Synchronized PulsAR Network (SPAN) technology is a system wide synchronization scheme that autonomously propagates a “heartbeat” information to all radios in a complex network resulting in the virtual elimination of self-generated interference.  This system wide approach, combined with the robustness of the pulsAR radios, delivers unprecedented jam resistant performance in the unlicensed band.

Interference Resistant Radio

In several of the pulsAR radio models we took the drastic approach of using a considerably narrower RF bandwidth than the majority of other devices operating in the unlicensed bands.  This brings several advantages, namely (i) the radio sensitivity is greatly improved allowing longer ranges, (ii) there is a much larger number of non-overlapping channels to choose from, and (iii) the radio is much more resilient to interference.  Refer to the pulsAR Wireless Ethernet Bridge data sheet for a list of other radio features specifically designed to combat interference.

The disadvantage of a reduced RF bandwidth is that the total throughput from an individual radio is also reduced.  However, the pulsAR-24027E for example, still delivers 2.4 Mbps to an individual user, and this is usually more than adequate in most applications.

Furthermore, at a system level, the SPAN technology actually allows you to deploy a central site with higher aggregate throughput than networks built with faster radios.

Complex Networks

As a network grows it becomes necessary to deploy multiple radios at the same site.  The reasons to co-locate radios include the following:

1. In a Point-to-Multipoint network you want to achieve 360 degree coverage around a central site, but would like to use sector antennas rather than one omni.  Sector antennas have higher gain than the omni and provide shielding from interfering signals originating at different sectors.  In this situation you might deploy a central site with six hub radios for example, each one feeding a sectorial antenna covering 60 degree sectors.
2. The number of remote radios serviced by a single hub has grown to a point where the shared bandwidth is no longer adequate.  You may then add a second hub radio operating on a different channel and split the remotes between two or more hubs.
3. You want to deploy a repeater site with two “back to back” radios.

The problem is that when you co-locate two or more radios they can become the source of self-interference, even if they are set to non-overlapping channels.  The reason for this is explained in the following section.

Co-located radios self-interference

When you co-locate two or more radios they can become the source of self-interference, even if they are set to non-overlapping channels.  This is illustrated in figure 1 below, that shows radio A transmitting on channel f1 while a co-located radio is trying to receive on channel f2.  Because the antennas are in close proximity, antenna B will pick up a significant portion of the signal transmitted by radio A.

Figure 1 also shows a block diagram of the radio front end circuitry.  It includes an RF filter to reject out- of-band signals, followed by a Low Noise Amplifier (LNA), a second RF filter, Mixer and finally the Intermediate Frequency (IF) filter.  Channel selection occurs at the Intermediate Frequency (IF), where the narrow band IF filter blocks out the other channels.  This means that if the interferer (radio A) is in close proximity, and is transmitting while radio B is trying to receive, it may saturate the LNA or the Mixer of radio B.  This results in radio B making errors even when it is set to a different channel than radio A.

The traditional approaches to reduce this self-interference include:

• Separate the antennas of the two radios further apart.
• Use different antenna polarizations.
• Lower the transmit power of the interfering radio.

Time Division Duplex (TDD)

The pulsAR radios operate in Time Division Duplex (TDD) mode whereby it transmits for part of the time and then receives for the remaining of the “cycle”.  What is unique about our TDD implementation is  that the cycle split is configurable:  you can choose between a fixed TDD cycle with a split configurable in 10% increments, or select an adaptive cycle where the radio chooses the split according to traffic patterns.

The fixed TDD cycle split is supported specifically to eliminate the self-interference problem described above.  By synchronizing the cycle times of all devices, co-located radios transmit at the same time and then receive at the same time.  This avoids the situation depicted in figure 1 altogether.  With a synchronized site you can then deploy upwards of 24 radios at the same location.

In a point-to-point topology the fixed TDD approach is straightforward.  With point to multipoint and linear network topologies, the fixed TDD scheme becomes somewhat more complex as the two phases may be split and used by multiple devices.

Refer to the Theory of Operation in the Operators Manual for a complete description of the fixed TDD scheme in point-to-multipoint mode.  The essential aspect however, is that with a fixed TDD scheme across all topologies, co-located radios are never in the situation depicted in figure 1.

SPAN Network Heartbeat

The key to synchronizing all the pulsAR radios in the network is the generation and distribution of the synchronization information or heartbeat.  This heartbeat must be ubiquitous such that all devices can pick it up.  In addition it must not depend on a single device for the generation of the heartbeat  since the whole network would fall apart if that device failed.

Figure 2 shows an example of a mixed network with multiple topologies, and illustrates the various ways in which the heartbeat flows.  When the whole network is synchronized, each radio runs its TDD in one of two phases, A or B, as shown in the figure.  All radios at a single site run on the same phase.

Propagation of heartbeat information occurs across the following mediums:

1. Over RF:  Allows a radio that assumed a master role to synchronize the remote radios.  The slave radios set their cycle timers to run out of phase from the master depending on the chosen fixed split.
2. Over Ethernet:  A unit may multicast a heartbeat packet to synchronize other co-located devices.  Devices that receive this heartbeat packet set their cycle timers to run in phase with the sender.
3. NetCrossing Gateway SYNC lines: If you use the Afar NetCrossing Gateways to carry serial traffic across the packet switch network, they are equipped with a dedicated SYNC connector through which they propagate the heartbeat without sharing the Ethernet WAN connection

SPAN Network Capacity

In a point to multipoint network, the SPAN synchronization scheme allows you to increase the capacity of your central site by simply adding more and more hub radios.  With 24 hub radios the aggregate throughput of the central site reaches 60 Mbps.

Synchronization is not possible with radios using the the adaptive TDD cycle approach (where the transmit and receive phases change dynamically).  Most manufacturers only support this adaptive cycle split because it has the advnatage of dynamically allocating the throuhgput to the direction with most traffic.  When you can not synchronize the transmissions of co-located radios you are usually limited to no more than three or four radios at the central site.  You would need radios with individual throughput in excess of  15 Mbps to achieve the same aggregate throughput.  Even then the narrow band nature of the PulsAR radios gives you numerous other advantages, including improved sensitivity, more non-overlapping channels, and better resilience to interference.

Conclusion

The self-interference problems created when you co-locate radios typically limits the number of radios at the same site to no more of three or four.  Even then, interference can only be overcome with special care during installation.  The Afar SPAN Network technology implements an autonomous synchronization scheme that ensures that all co-located radios transmit and receive at the same time.  The synchronization scheme, or heartbeat, is simple to configure and spreads through complex networks so that multiple sites with co-located radios are all automatically synchronized.  In a SPAN network, when you take advantage of the large number of non-overlapping channels offered by the PulsAR-24027 radio, upwards of 24 radios can be co-located without self-interference.   This translates into a higher central site capacity than what is typically possible, even with higher bandwidth radios.