Designing Resilient Connectivity for Continuous Water Monitoring

Unlike manual sampling at defined intervals, continuous monitoring uses unattended sensors to take regular measurements and make those readings available to the systems that record them or act on them. In drinking water, online monitoring can include parameters such as turbidity, pH and chlorine residual. Sensors also require calibration, validation and maintenance if their readings are to support operational functions such as alarms, automated shutdowns and the disinfection contact time calculation.

Two separate monitoring regimes are driving the deployment of continuous water-quality monitoring. In drinking-water supply, the Drinking Water Inspectorate can require continuous monitoring through a notice under the Water Supply (Water Quality) Regulations. For receiving water, including rivers and other watercourses affected by wastewater discharges, section 82 of the Environment Act 2021 establishes the framework for monitoring upstream and downstream of storm overflows and sewage disposal works. The regulations, assets and monitoring requirements differ, but both regimes create the same fundamental connectivity question: how will readings travel reliably from the monitoring position to the systems that record them or act on them?

Monitoring positions are determined by the water system rather than by mobile coverage. A service reservoir may occupy elevated ground, a network chamber may be below ground and battery-powered, and a raw-water monitoring point may be remote from mains power. Mobile coverage, power availability and the required reporting interval therefore need to be assessed at the installed position. A common architecture and assurance approach can be applied across both monitoring regimes, with the implementation matched to the conditions at each monitoring point rather than based on a general assessment of the wider site.

Drinking water supply

Under regulation 27 of the Water Supply (Water Quality) Regulations, every treatment works and connected supply system needs a risk assessment that is kept under continuous review. Each company must report every assessment and review under regulation 28(1). Where a report identifies a significant risk, the Inspectorate can serve a notice under regulation 28(4) requiring specified measures within defined timescales. In 2025, the Inspectorate served 99 legal instruments in England, of which 87 were regulation 28(4) notices.

A regulation 28(4) notice can specify what must be monitored and what the online monitoring system must be connected to, and published notices have required continuous monitoring with alarms and fail-safe shutdown. Each notice specifies the measures required and the dates that apply.

The engineering implication is that, where a monitoring or control function depends on connectivity, loss of that connection can prevent the function from operating as intended. The end-to-end reporting path should therefore be documented and tested alongside the monitoring system, rather than treated as a separate connectivity decision after the sensor has been selected.

Receiving water

Section 82 of the Environment Act 2021 creates the statutory framework for continuous receiving-water monitoring. The substantive monitoring duty in section 141DB of the Water Industry Act 1991 has not yet been commenced. Since 3 November 2023, section 82(1) has been in force only for the purpose of making regulations under section 141DB. When the duty is commenced, it will apply to sewerage undertakers whose area is wholly or mainly in England and require continuous monitoring upstream and downstream of storm overflows and sewage disposal works that discharge into a watercourse. Defra’s programme guidance, as updated by its September 2023 consultation response, sets out the current technical and rollout expectations described below.

• What is measured. Section 141DB names dissolved oxygen, temperature, pH, turbidity and ammonia, and allows regulations to specify additional parameters.
• Where. One monitor upstream and one downstream of each asset. Outlets within up to 1 km of one another in a single length of watercourse can be monitored as one cluster, capped at ten.
• By when. Defra’s September 2023 consultation response set an initial rollout focused on high-priority sites, with 25 per cent delivered by 2030 and the timetable reviewed in 2027.
• How often. At least hourly, switching to at least every 15 minutes during storm overflow operation or a final effluent quality event, and staying there for 24 hours afterwards.
• When it must be published. Within an hour.

The programme’s reporting and telemetry expectations bring connectivity directly into the monitoring design. Defra’s provisional technical guidance expects each monitor to carry telemetry that allows the interval to be switched, manually and from contextual data such as weather radar. Defra’s interim technical standard requires remote fault reporting. The recommendations appended to it add an internal buffer of at least 3,000 readings and ask sewerage undertakers to make clear when monitors or telemetry services are not operating.

With a number of cellular network operators available in the UK, a multi-network connectivity service can provide access to several mobile radio networks at a remote site. This enables a remote transmission gateway to have access to multiple networks if they are available at the location. While this approach can increase the connectivity options available at the monitoring position, it addresses only part of the resilience requirement, as a multi-network service can still retain a shared mobile-core dependency. Access to multiple local radio networks does not, by itself, establish operator-core independence. Where a site can tolerate delayed transmission, radio-network diversity may provide an appropriate level of resilience. Where a reporting interval, one-hour publication requirement or remote alarm function cannot tolerate a shared-core outage, the assessment should consider whether an independently routed alternative path is required. That decision needs to be made for each installed position.

A resilient connectivity design should distinguish radio-network diversity from operator-core independence. Where the operational requirement calls for independently routed connectivity, the architecture can use two independent mobile operator-core paths, switching logic within the connectivity equipment, and private network infrastructure with encrypted connections on both routes. The resilience achieved in practice will depend on the selected operator profiles, connectivity services, equipment configuration and communications bearers.

Users of the Openreach public switched telephone network (PSTN) must be migrated to digital services by 31 January 2027, while the UK’s mobile operators plan to retire their 2G networks between 2029 and 2030. A monitoring point that depends solely on the PSTN or a 2G mobile connection will lose its monitoring data reporting path when the service it uses is withdrawn, so an alternative network must be assessed for its performance and resilience, commissioned and tested beforehand. The government’s 2G Switch-off Charter includes commitments intended to protect Critical National Infrastructure during the transition. It also makes clear that organisations using 2G remain responsible for identifying affected services and taking appropriate steps to maintain connectivity. Water companies should therefore identify any CNI assets that continue to depend on 2G and plan their migration before the relevant network is retired.

Replacing PSTN- or 2G-dependent monitoring and telemetry equipment requires device and communications specifications to be reviewed. Resilience, security, failover behaviour and acceptance-test evidence should be defined at that stage, before the replacement design is fixed. Addressing these requirements early can reduce the risk of redesign and repeat site work later.

A monitoring point depends on its reporting path for readings to reach the systems that record them or act on them. The availability and resilience required from that path will depend on the reporting interval, the operational purpose of the data and the consequences of delayed transmission.

Published evidence provides useful context on communications coverage and network disruption. Ofcom recorded 616 reported resilience incidents in the year to August 2025, affecting 12.7 million customers and accounting for approximately 192 million customer hours of lost service. These figures demonstrate the wider operational relevance of network resilience, but they do not quantify water-telemetry availability or establish a connection between communications loss and a water-quality event. Ofcom also does not categorise the reported incidents as core or edge failures. Operator-core dependency therefore needs to be assessed through the proposed architecture rather than inferred from national incident totals.

National coverage figures provide useful context, but they cannot replace an assessment at the proposed monitoring position. UK 4G geographic coverage reaches 96.2 per cent of landmass from at least one operator and 83.6 per cent from all four. In rural areas, the corresponding figures are 95.7 and 81.7 per cent, while 4.3 per cent of rural landmass has no predicted 4G coverage from any operator. These figures are based on operator signal-strength predictions rather than measurements at an individual monitoring point. National statistics can inform the design, but they cannot establish the coverage or resilience available at the installed position.

Continuous water-quality monitoring brings the sensor, telemetry, connectivity and receiving systems into one operational path. The connectivity specification should therefore establish:

• the communications dependencies between the monitoring point and the systems that record the readings or act on them;
• whether the design provides radio-network diversity, operator-core independence or both;
• the required behaviour when the primary connection becomes unavailable;
• how readings are buffered, protected and forwarded during and after an interruption;
• how connectivity faults are identified and reported;
• the residual exposure if individual communications services or shared infrastructure become unavailable; and
• the evidence required at factory acceptance testing, site acceptance testing and handover.

These requirements should be defined before equipment and connectivity services are selected. They can then be tested against the conditions at each installed position, including local power availability, mobile coverage, the required reporting interval and the operational consequence of delayed data.

The following case studies show CSL’s experience of large, distributed water-sector estates, telemetry migration and partner-led integration.

Thames Water, with Vodafone Business. CSL and Vodafone Business delivered resilient connectivity for 3,500 telemetry sites as part of Thames Water’s migration away from the PSTN, demonstrating experience of large-scale water-sector telemetry transformation. Read the Thames Water case study.

Severn Trent, with SecuriPlex. CSL designed and delivered secure cellular connectivity for security systems across hundreds of remote, often unmanned sites. The architecture used dual-SIM resilience and encrypted VPN tunnels segregated from the public internet, while SecuriPlex managed the site integration. Read the Severn Trent case study.

CSL can assess the connectivity dependencies and conditions at each monitoring position, identify appropriate implementation routes, and provide an architecture note with an acceptance-test evidence plan. Enquiries to sales@csl-group.com.