Communication options for distributed M2M systems
This is a comparison of the various technology options currently available when designing and deploying a distributed M2M communication system.
Determining the suitability of a technology for a particular application requires consideration of the strengths and weaknesses of that technology, as well as the specific requirements of the application (such as distance, terrain or bandwidth required). Additionally, their cost structures can be quite different, meaning that consideration of their long term viability (as well as initial setup costs) is required.
Figure 1 – Setup costs vs number of stations for competing communications technologies
(not inclusive of per-metre installation costs for ADSL/Fibre/Copper)
Fibre Optic Cable
There are several advantages to using Optical Fibre. It provides a reliable communications link with wide bandwidth and is impervious to electrical interference, making separation between power and communications conduits unnecessary. The hardware is readily available from a wide range of vendors, and the technology has a long life with minimal running costs.
On the downside, the costs of deployment are very high. The cable must be laid underground, typically in conduit – such that the cost of trenching alone makes it uneconomical for long distances. Also, the more affordable Multi-Mode Fibre (e.g. OM1) is limited to runs of less than 5km.
Given the implementation cost, and the relatively low bandwidth requirements for distributed SCADA system, fibre optic cable is unlikely to be a viable option for this application. While it may be appropriate in certain circumstances for point-to-point communications within a site (especially when bandwidth requirements are high), it is definitely not appropriate for use between sites.
Copper communications cable (Ethernet, RS485)
There are several advantages to using traditional copper cabling, especially for connecting sub-systems within a site. It is a reliable and mature technology, providing a reliable and physically independent secure network with almost no ongoing costs. Bandwidth is relatively low compared to optical fibre (especially for RS485) and distances are limited (100m for Ethernet vs 1.2km for RS485). Disadvantages include the costs of laying cable underground, the susceptibility to water damage and electrical interference, and the need for surge protection between sites.
Copper cable will be an important element of our communication networks for a long time to come, however it is far from an economic solution for long distances, so its use will be mostly for short runs where the bandwidth of optical fibre is not required.
ADSL Broadband can connect each of the main sites to the private network on the internet.
This is a well established technology with hardware readily available from many vendors. The bandwidth is relatively high (although asymmetric, with much slower uplink speeds) and the cost per data quota allowance is less than 3G or other mobile technologies.
On the downside, it is likely to be expensive to implement in remote locations due to the cabling requirement (which also needs to be separated from power cabling to reduce interference). In addition, the cost structure does include a monthly fee, unlike technologies such as digital radio.
Digital Radio Links
IP radio technology provides a solution that is suitable for both long and short distances. The installation and deployment costs are not trivial (due to the aerials and masts required, as well as radio surveys & licensing fees). Also, there are distance limitations dependent upon terrain, height of masts, radio power etc. Bandwidth is relatively low, although this is probably not an issue for most SCADA applications.
Naturally, while it is more prone to interference than technologies like optical fibre, it is also less susceptible to some outside influences (such as power failure) than 3G-based solutions. With appropriate UPS backup at both ends, digital radio can provide reliable communication during power outages. On the downside, the aerials and masts are susceptible to vandalism and lightning strikes.
The ongoing costs are relatively low, but there is an annual licensing fee – so this technology isn’t quite as cheap to run as fibre or copper cable. The relatively high cost of the initial hardware setup also means that allowance for periodic replacement of hardware will be a significant proportion of the ongoing costs.
Telstra 3G Data Network
Telstra, and other carriers, offer private machine to machine networks built on top of their public mobile phone networks.
There are several advantages to this technology – it provides a secure and reliable network that is cheap and easy to implement. It is effectively distance independent, the network could cover the entire country with no impact on the deployment cost. The hardware is readily available from many vendors, and the installation is simple and low profile (not requiring the large aerials or masts needed for digital radio).
On the downside, the shared bandwidth means that network latency can vary depending on the level of network congestion. Network availability can also be adversely affected by other external conditions such as power failures. Also, the cost structure of 3G solutions offsets the low setup costs with ongoing monthly fees.
Additionally, there are stiff financial penalties for exceeding data usage quotas, so this aspect of system planning needs to be treated with care. It may be worth undertaking a survey of data usage by deploying a packet sniffer/protocol analyser to gather hard data on your precise data usage prior to selecting a data plan.(Fine tuning of low-level network parameters such as TCP KEEPALIVE packet timing and periodic PINGs between devices can yield significant reduction in data overhead, especially on a lightly-loaded network.)
Choosing an option
Several options (fibre, copper) are not suitable for long distance runs due to their excessive capital costs. Of the two options (3G, IP radio link) which are suitable for communications between remote sites over relatively long distances, their cost structures are somewhat different. Digital radio has a high setup cost, but relatively low running costs (basically the annual licensing fee, plus an allowance for periodic replacement of hardware). For a small number of stations, the extra setup costs might be offset by the lack of ongoing costs, while a more extensive network is unlikely to recoup the extra capital costs in savings over the lifetime of the network.
The cost structure of 3G has relatively low initial costs, while the fixed running costs of the network are somewhat higher than digital radio. This makes it a more attractive option when the number of stations on the network is sufficient to offset this.
Typically, if the number of stations exceeds around 10 or so, the 3G solution will retain its cost advantage vs. the equivalent digital radio system over the long term.
Figure 2 - Combined Setup and Operating costs over 10 years
The hardware price differential between 3G and ad-hoc digital radio networking is likely to increase rather than decrease over time, reflecting the fact that the 3G infrastructure used for M2M applications is shared with popular consumer-oriented products, leading to a highly competitive market for the supporting semiconductor technology, and the resultant economies-of-scale. Digital radio is, and is likely to remain, a relatively niche product with hardware costs reflecting this.
One of the downsides to utilizing consumer-oriented infrastructure such as the 3G network is the relatively rapid rate of technological obsolescence - digital radio systems can (barring changes in government regulation of the RF spectrum, and/or availability of replacement hardware) remain in service almost indefinitely, while a mobile-network based solution will require updating as the carriers eventually retire obsolete systems. This may well be offset by the lower cost of the mobile-network based hardware, however.
While IP Private Mobile Radio networking is relatively expensive when deployed as the backbone of a large scale network, it may prove to be a useful adjunct to a 3G-based network – providing network access where cellular coverage is limited, or where immunity from network congestion is a requirement. The ‘network agnostic’ nature of most sensors and RTUs makes this mixed technology approach a viable option.
Carrier-hosted virtual server
A useful strategy which makes effective use of the private network provided by the 3G carrier is the hosting of SCADA systems on a virtual server in the carrier’s own datacenter. While there are ongoing fees for such hosting (especially with regard to the virtual server’s disk capacity – which may require careful management) there are also a number of significant advantages.
Connection to the M2M private network is relatively straightforward, while the SCADA system can also host services facing the internet (such as email notification or web-based UI) as required.
Backup, redundancy and disaster-recovery resources are provided by the carrier – this is likely to be a significant advantage relative to having the SCADA system hosted by the end-user, where such resources may be inadequate.
For large-scale systems with multiple stakeholders, it relieves the burden of SCADA hosting from any single stakeholder and provides equality of access (via VPN or web-based UI) to all of the systems’ users.
The selection of communication technologies is a complex decision, with no simple “winner” being appropriate for all situations. That said, there is a strong case for implementing a 3G M2M network as the “backbone” between remote sites, especially for systems with a number of sites spread over a wide geographic area. Likewise, there are a number of unique advantages to other technologies such as IP radio, and it may be advantageous to apply them to appropriate parts of the system.