Mesh wireless sensor networks: Choosing the appropriate technology
While most network approaches use routing as the basic architecture, new flooding-based technology offers distinct advantages, especially when it comes to larger networks.
Selecting an appropriate technology from a range of possibilities is the key to any successful system design, and selecting a Wireless Sensor Network (WSN) technology is certainly no different.
The two main technologies used to transmit data in mesh-based WSNs are flooding and routing. Flooding is also used in routing-based WSNs to enable the route discovery process.
New flooding and routing developments are showing up in the market today. A look at both technologies’ advantages and disadvantages in various types of WSNs, particularly those used in smart grid and large-scale smart metering networks, can help network designers make more informed decisions.
Routing in WSNs
Routing in wireless networks has been an active area of R&D for many years. Routing techniques rooted in computer data communications have been thoroughly explored for use in wireless networks, resulting in the emergence of many self-organizing, self-healing models in commercial implementations.
The reason for all this activity is that robust operation within changing propagation conditions and under energy and communication bandwidth constraints precludes the use of traditional IP-based protocols and creates a difficult challenge for dedicated WSN routing algorithms. The task of finding and maintaining routes in WSNs is nontrivial because energy restrictions and sudden changes in node status (including failure, jamming, or temporary obstructions) cause frequent and unpredictable changes. Building and propagating automatic routing through the network requires powerful node processors, large amounts of memory, and additional dedicated routers, as well as network downtime until alternative routing is established.
Building and maintaining routing tables with alternate routing (for responding to changing propagation conditions) while using low-cost, low-power processors proves to be a formidable challenge, which is amplified when the size and number of hops increase.
Many new and sophisticated algorithms have been proposed to resolve these issues. The resulting routing schemes take into consideration the inherent features of WSNs along with application and architecture requirements. To minimize energy consumption, routing techniques employ some interesting techniques special to WSNs, such as data aggregation and in-network processing, clustering, different node role assignment, and data-centric methods.
These routing techniques seek balance between simple solutions with limited robustness and sophisticated solutions. Even in sophisticated solutions, there is still the risk that in large networks or when messages are short, the routing overhead will consume valuable resources such as bandwidth and power and sometimes cause packet collisions. Worst case, these factors combine to finally degrade network robustness, throughput, and end-to-end delay.
One basic routing attribute related to the dynamic nature of an RF environment has yet to be solved: One moment after the routing table is created, it is already obsolete because the RF conditions have changed.
Flooding the network
In flooding, instead of using a specific route for sending a message from one node to another, the message is sent to all the nodes in the network, including those to whom it was not intended.
The attractiveness of the flooding technology lies in its high reliability and utter simplicity. There is no need for sophisticated routing techniques since there is no routing. No routing means no network management, no need for self-discovery, no need for self-repair, and, because the message is the payload, no overhead for conveying routing tables or routing information.
Flooding technology has additional advantages related to propagation. Signals arriving at each node through several propagation paths benefit from the inherent space diversity, thus maximizing the network robustness of handling obstructions, interferences, and resistance to multipath fading, with practically no single point of failure. In other words, blocking one path or even a limited number of paths is usually of no consequence.
Furthermore, no routing means that the controller is extremely simple, requiring minimal computing power and memory and thus low power consumption, low PCB real estate, and low cost.
Despite these benefits, flooding the network with repeated messages has its own challenges, precluding the use of flooding until a short time ago, except for route discovery in routing-based networks. For transmitting data, the main questions are how collisions are avoided, how the retransmitting process propagates the message efficiently toward its destination, and how the process ends, without an energy-wasting avalanche.
Fortunately, a novel flooding approach using a synergic combination of techniques enables designers to answer these questions and solve the challenges. Incorporating time division multiple access combined with high-accuracy synchronization allows the retransmissions to occur simultaneously so that the message propagates one hop in all directions at precisely the same time and avoids collisions. At each hop, nodes retransmit only relevant information, and the number of retransmissions corresponds to the number of hops in the network, so there is no waste of retransmissions.
Figures 1 and 2 exemplify the different propagation patterns for the two mesh technologies in the same 24-node, three-hop network. The first hop signal propagation is blue, the second is purple, and the third is green.
In Figure 1, the effect of a signal obstruction or interference in the route on the left precludes the signal from arriving at its destination. Sophisticated routing-based schemes identify the problem and try to reroute the signal. If no alternative route helps, a new routing is recalculated, leading to side effects such as latency and possible service interruption until the new route it completed, verified, and propagated.
In the flooding-based scheme in Figure 2, a signal obstruction will most likely not affect the operation at all because of the numerous redundant paths.
Challenges and design issues
WSNs have several inherent restrictions compared to IP networks. The main restrictions are: limited bandwidth of the wireless links connecting sensor nodes, limited computing power, and limited energy supply, especially in case of battery-operated networks. One of the main design goals of WSNs, especially in smart grid and smart metering, is to provide highly reliable and continuous communication of all the nodes in the network, all the time under the changing propagation conditions.
This can be notably challenging when operating in the unlicensed ISM frequency band. The propagation changes can be caused by obstructions such as a large metallic object in the line of sight of two or more nodes, by changes in the nodes’ location, and especially by man-made sources of in-band noise. Designers must consider the challenges and design issues when using the two WSN techniques under these restrictions.
The two types of networks handle network deployment and maintenance differently. Routing-based networks employ special techniques to self-organize by discovering the network nodes, finding the optimal routing table, propagating it through the network, and continuously changing it according to changeable propagation conditions and physical network changes. These techniques, some of which are sophisticated, operate well when the networks are small and simple. However, when the networks are larger with large numbers of nodes and hops, then expert human intervention and dedicated management software are usually needed for both initializing the network and coping with major changes. Furthermore, until the network reorganizes, parts of it or the whole thing can be down for various periods of time.
In the novel approach of flooding-based networks, this design challenge does not exist because these networks always use all the available propagation paths. This precludes the need for self-organizing or human intervention, and large networks are handled with the same simplicity as small ones.
The same principle applies to a related challenge: scalability. Increasing the size of the network increases the complexity of the routing table and accordingly, the probability of network failure with routing. In flooding-based networks, the same network size increase causes the network to be more robust. This issue is important in smart metering because the Advanced Metering Infrastructure (AMI) must be highly reliable.
Energy consumption is one of the most important WSN design challenges. In routing-based networks, the total number of operating nodes at any moment (when the network is transmitting) is always lower than in flooding-based networks; thus, routing consumes less energy. On the other hand, flooding-based messages are much more efficient, as they do not need the overhead associated with transmitting routing tables and commands, which increases with the number of nodes and hops. In modern flooding-based systems, the energy of the signals received from adjacent nodes adds up, so less power can be used for achieving the same range.
For conclusively determining the best approach, energy efficiency depends on the specific WSN application, network size, and message size. Therefore, network designers should compare the two types of networks, taking into account the current and expected parameters.
As a rule of thumb, routing technologies are expected to have lower energy consumption in smaller networks with very few hops. Flooding-based technologies are expected to have lower energy consumption in large networks with many nodes or hops, or when the average size of the message is small.
Range and coverage are among the most important WSN design challenges. In networks with simple topology, the range of the network is directly related to the range between two adjacent nodes and thus is affected by the quality of the physical layer circuitry and software. In routing-based networks using mesh topology, the network range also depends on the routing quality of each single path between each two elements of the network. In flooding-based networks using mesh topology, the nodes sum up the energy from all the received nodes, creating a much better range of the whole network compared to the most sophisticated routing network. In addition, multiple propagation paths improve network coverage to a “no dead spots” quality.
Latency importance in WSN design depends heavily on the specific type of application. For example, in an irrigation application where the time resolution for water pump switching is tens of minutes, latency is of no consequence. On the contrary, in a utility application where customers call the utility staff about current meter reading, latency determines the quality of the staff service. In some industrial control applications, a delayed feedback to an emergency situation can be very costly or even perilous.
When comparing the two technologies, flooding-based networks with mesh topology have an inherent advantage because of their lower overhead. Furthermore, routing-based networks suffer from latency inconsistency caused by possible RF propagation problems in the designated route, an inconsistency that increases with the number of hops. In flooding-based networks, propagation problems in one route usually do not affect latency at all because of the inherent space diversity.
Simple, route; larger, flood
Designing a modern WSN is a challenging mission due to the wide selection of technologies available in the market. A methodical approach starts with the choice of topology. Today, there is no doubt that mesh networking provides the highest-performance and most flexible topology in practically all WSN applications, except the most simplistic ones. The next important choice is between a routing- or flooding-based network. Here, the selection depends on the specific application and its scale.
In applications that require simple networks with a small number of nodes and hops, routing-based networks can be the best option. A typical application is a home area network, where the number of nodes is limited and operation reliability is not paramount. On the other hand, in networks with a large number of nodes and hops, especially where robustness and reliability are important, flooding technology provides a more suitable solution. Typical applications with these larger node counts are smart metering and highway illumination.
Leor Hardy leads the R&D team at Virtual Extension (Givatayim, Israel), which pioneered advanced flooding technology, brought it to maturity, and implemented it in various projects such as smart electrical metering for Israel Electric Company and smart water metering for the Mamilla quarter in Jerusalem. In his previous positions, Leor participated in several R&D projects for TeleSciCOM, Nexus, and Elisra, and served as a senior consultant to Alvarion and Metalink. He received his B.Sc. in Electrical Engineering from the Technion Institute of Technology.
Marius Gafen is a senior researcher at Virtual Extension. In his previous positions, Marius led several R&D projects for MoD and private high-tech companies in Israel, including ArelNET, NSIcom, Coresma, and Sonarics. He received his B.Sc. in Electrical Engineering from the Technion Institute of Technology.