# Not a Bridge Too Far: Connecting IoT Devices

### Key Takeaways

• The major topologies available for IoT design.

• A discussion of the importance of (de)centralization to a network.

• At-a-glance comparison between the basic IoT topologies.

Connected IoT devices can trace their roots back to a famous mathematical problem.

The Seven Bridges of Königsberg is a fundamental mathematics problem that was proven to have no solution by the formidable Leonard Euler. The problem asks if it’s possible to cross these 7 bridges sequentially on a single walk throughout the city and its adjoined islands. Unbeknownst to Euler at the time, he was laying the groundwork for graph theory, a discipline of mathematics that would later lead to the study of topology.

Amazingly, the methods of connecting IoT devices rely heavily on the mathematics originally developed nearly 400 years ago. This knowledge can be supplemented with developments in electronics and technology to help elucidate the best layout available for networks depending on what performance metrics need to be minimized or maximized. To begin, some of the most common topologies are presented to understand the reasoning behind different network configurations.

IoT systems can be reduced to three complementary layers: edge sensing, gateways, and data systems. Each of these three groupings performs specific functions within the system, but IoT designs can be viewed more abstractly as a collection of nodes and links. Imagine a set of wooden sticks and spools popular as children’s toys: the node represents the point of interest or connection point, whereas the links are used to directly tie adjacent nodes together. These two building blocks are used to construct a variety of different network topologies. The topologies represent different network configurations that are built to extract the best performance in the most efficient method possible:

• Point-to-point - The simplest network, consisting of a link between two nodes. A two thing-network is unlikely to see much use in an IoT setting due to the limited scope of communication, but a dynamic point-to-point design (which allows for switching between targeted nodes) can offer solutions in some instances. Overall, however, the point-to-point model heavily constrains IoT, which generally tends to favor broadcasting information instead of singular one-way or bidirectional connections.
• Bus - Eminently familiar to layout designers, the bus topology uses a central transmission line to which all nodes are connected. This ensures the message travels to all nodes in the network (barring interruptions), but two major limitations of the network are the physical connectivity of the network (reduces range and scalability) and the lack of redundancy in the case of cable failure.
• Daisy-chain - All nodes are connected in series, which can take two forms. The first, linear, is less useful due to the directionality: in the worst-case scenario, a signal has to pass through the entire length of the network if going from endpoint to endpoint. A much more valuable implementation is the ring variant, where the endpoints are connected to one another. This not only reduces travel time across the network by half for what would be the longest linear walk, but the reliability of the network improves as a single interruption between two nodes, effectively returning the network to a linear daisy chain.
• Star - The star topology ties all periphery nodes to a central node, which acts as the nexus for communication. Since each additional periphery node only adds one connection, star networks are highly scalable, and the lack of direct pathing between periphery nodes is a security feature. On the downside, the high load experienced at the hub of the network can cause congestion.
• Mesh - A full mesh network ties every node to every other node; communication can jump between any two points in the network in the fastest possible pathing available. This saturated network has the advantage that an interruption to any one link can be surmounted through more circuitous pathing between nodes, making this network style highly reliable. However, this topology becomes highly strained for large networks, making a true mesh unwieldy once the total number of nodes reaches an appreciable amount. Furthermore, meshes need to be distributed densely across a relatively short distance to maintain the signal strength between the nodes, but this is made up for with low power consumption and the ability to disseminate data rapidly throughout the network. Partial variations of the mesh help to circumvent some of the drawbacks of the full mesh layout, focusing on connecting critical nodes or bypassing long walks in a similar setup to the star layout.

Topologies can also be hybridized by combining the fundamental layout of two styles. These include trees (individual star networks connected by a central bus), star-rings (a ring formed between the central node of individual star networks), and linear daisy-chained stars. These can also include self-hybridized networks like a distributed bus (a bus of buses) or a star of stars. There are even distribution methods that emphasize connections over nodes, such as a grid layout, which is a superset of daisy-chains and can be further generalized to higher dimensions. In dimensions greater than one, the network need not be saturated with nodes at every intersection of connections; in other words, each node has at least two adjacent neighbors, though the walking distance between neighbors is variable.

## The Cost-Value Analysis of Centralization

Topologies can also be evaluated based on the level of centralization, or decentralization, contained within. On the extreme ends, a full mesh network represents a perfectly decentralized network, while a star network indicates complete centralization. The centralization of most networks exists on a spectrum between these two cases.

The countervailing forces guiding the connectivity include network redundancy and scalability. The star network represents zero redundancy: damage or interruption to the signal between nodes results in the isolation of that particular path. More worrying still is that the hub represents a single failure point of the network; a highly integrated node relative to its neighbors represents efficiency in connectivity, but a liability in redundancy. On the other hand, the number of connections between nodes in fully decentralized networks grows quadratically, and can quickly become unwieldy even in IoT networks of moderate size. System designers need to balance the robustness of connectivity against constraints.

 Topology Related Pros Cons Centralization Point-to-point N/A Simple Direct Limited IoT application N/A Bus Tree Distributed Broadcast to all nodes Single point of failure High Daisy-chain Ring Linear Grid Good redundancy (ring) Improved pathing (ring) Increased walk path (linear) Reliability (linear) Low Star Tree Star-ring Minimizes redundancy for cost efficiency Single point of failure High bandwidth at the central node Maximum Mesh Full Partial High redundancy for improved reliability Cost Range Design complexity Minimum (full) Low (partial)

## Harnessing Communication in Product and Toolset

Connecting IoT devices requires some forethought as to how data flows through the network: what are the most important nodes and what layout best serves the needs of the system? It is not a simple question to answer, and with technology evolving at a rapid pace, so too must system design. Supporting this requires flexibility for the engineering, layout, and networking teams to produce the best possible solution.

From proof of concept to production, Cadence offers extensive PCB design and analysis software that enables teams to rapidly simulate and layout boards without sacrificing quality in design. With OrCAD PCB designer, layout designers have a potent mixture of functionality, power, and ease of use for IoT or any other cutting-edge board style.

Leading electronics providers rely on Cadence products to optimize power, space, and energy needs for a wide variety of market applications. To learn more about our innovative solutions, talk to our team of experts or subscribe to our YouTube channel.