Bulletproof Wireless Reliability with Redundant Links

Imagine that two generals are each camped with their armies on two hilltops, with enemy forces filling the valley in between. The first general sends her best scout to sneak through the valley and deliver this message to her counterpart: "Our only hope is a coordinated assault—a mistimed attack would be disastrous. If I have been assured that you will join me, I will attack at midnight tonight." The second general responds by sending a messenger with a reply to the same effect.

The problem is that even if both messages get through, neither general could have absolute confidence in the attack. Due to the unreliability of their communications, the second general does not know if his reply was received, so he does not know how the first general will act. Meanwhile, even as she holds the reply in her hands the first general knows that the second general will be uncertain about the status of his reply. This uncertainty compromises the confidence that both generals need to attack.

The startling truth is that no matter how many messages the two generals send neither can ever be absolutely confident in the attack, because the status of the last message in the exchange will always be in doubt. This dilemma, known as the "two generals' problem," exists any time communications reliability is uncertain, and illustrates the importance of finding strategies to increase the reliability and decrease the unpredictability of a communications system.

Industrial automation may not appear to have the gravity of armies and generals, but coordinated activity is just as crucial to successful management of an intricate industrial process. More and more, system integrators are turning to wireless communications as a coordinated solution for highly distributed or mobile applications, such as massive stacker reclaimers or underground mine cart systems. Common to these systems are constantly moving components, whether they are mine carts or the arms of cranes, which make wired networking impractical. However, reliable communications haven't become any less important just because you have switched to a wireless system. A network interruption can cause costly downtime and expensive damage, or even threaten the safety of your workforce.

In order to create a reliable network, we have to recognize the various challenges of wireless communications. Even in a properly configured wireless network, there are two primary sources of communications errors: collision and weak signal.

Collisions occur when multiple devices compete simultaneously on the same network segment. For wired Ethernet, transmitting stations can listen for incoming signals and notify the network if a collision is detected. Confident collision detection is rarely possible in wireless networks. While transmitting, a wireless station will most likely receive only the signal from its own transmitter, which is located closest to its corresponding receiver. This obscures any collisions from other stations. As a consequence, if packet loss is detected in a wireless network, the 802.11 protocol assumes that a collision occurs, and responds with exponential backoff and adjustments of the retry counts.

In addition, packet loss can occur if the signal strength at the receiver is too weak for the data rate of the packet. Weak signals can be caused when the network attempts to transmit at a data rate that is too high for the signal strength, or because physical obstacles are making the signal weaker. Simple movement could also be the culprit in mobile applications if the client has moved farther away from the access points. The 802.11 protocol only begins to tune the data rate and power, or consider AP hand-offs after a collision has been ruled out as the cause of packet loss.

Environmental interference also contributes to wireless unreliability. Electronics such as cordless telephones, Bluetooth devices, industrial microwaves or video senders all create interference on the 2.4 GHz band used in most 802.11 wireless networks. This co-channeling and radio interference reduces band availability and also causes more error packets.

Redundancy is a common strategy to increase the reliability of a system, including wireless networks. At its simplest, redundancy is simply the technological application of a basic concept—always have a spare. Redundancy can be implemented on many levels. Power redundancy, which provides devices with a backup power supply, is fairly common and relatively simple to implement, while complete system redundancy, which provides a top-to-bottom duplicate of the entire system, is extraordinarily expensive and frequently impractical.

Creating a redundant wireless link does add a layer of redundancy to the system and increase the reliability. While one wireless link is active and transmitting, the other is on standby on a different channel, ready to take over if transmission quality drops below a certain threshold. However, this solution has its limits. First, it takes time to switch to the new link. Some packets will invariably be lost during this transition period. Second, if the threshold for activation of the standby link is set too low, then communications will need to drop to a low data rate before the current link is disconnected. This is unacceptable for applications that require a continuous and high level of performance. However, if the threshold is set higher to avoid this very problem, the "ping-pong" effect—where the wireless connection constantly switches back and forth between the two links—becomes a potential problem, dramatically increasing the complexity of the switching mechanism.

In order to avoid these complications, Moxa's wireless products add advanced refinements to redundant wireless link technology. The AWK-5222/6222 features two independent RF modules. This means that both modules can be activated simultaneously. By working concurrently, the redundant wireless link duplicates the data transmissions, which eliminates any packet loss from waiting for the connection to switch links. The ability to broadcast simultaneously on two distinct channels, and even two distinct bands (2.4 GHz and 5 GHz), also bypasses any interference that might exist in one band.

Simultaneous dual redundant links also increases the throughput by reducing the number of resend requests. Throughput is the average rate with which messages—the data that the user is actually interested in—are delivered over a path. It's possible to have high data rate yet unsatisfactory throughput if a lot of communications "overhead" is consumed on information such as frame headers and retransmission, leaving little room for actual message data to be delivered. Resend requests, which are a particularly prodigious source of overhead in wireless applications, can adversely affect throughput. Fortunately, with dual redundant links, data packets do not need to be resent as long as they are received on one of the two links. This minimizes communications overhead, optimizes bandwidth utilization, and maximizes throughput at any data rate and under any conditions.

The quality of wireless communications can also be compromised by equipment failure or a simple repositioning of the client's physical application space. Moxa's redundant clients are sufficiently flexible and adaptable to respond promptly to device failures or movement relative to APs, and to prevent packets from being lost. Moxa's redundant clients combine Turbo Roaming technology and redundant wireless technologies to proactively search for APs in concert. When wireless performance drops below a certain threshold, the client will identify an optimal AP and switch immediately instead of waiting for the existing AP signal to be completely lost.

» Back to index