A mobile network is not a static infrastructure — it is a continuously evolving, dynamically reconfiguring system that responds in real time to millions of simultaneous events. Understanding what happens inside that system across a typical 24-hour period reveals the remarkable engineering complexity that underpins what users experience as simple, seamless internet connectivity.
The Morning Peak: When Millions Attach Simultaneously
Between approximately 7:00 and 9:00 AM on a typical weekday, a mobile network experiences one of its most demanding operational periods: the morning attach surge. As commuters power on devices, leave home Wi-Fi networks, and transition to mobile data, the network must process an enormous volume of simultaneous Attach Requests, PDN Connectivity Requests, and data session initiations within a short timeframe.
A single MME (Mobility Management Entity) in a large urban operator network may process tens of thousands of attach procedures per second during peak periods. Each attach involves multiple signalling messages exchanged between the UE, eNodeB, MME, HSS, S-GW, and P-GW — with cryptographic authentication, policy retrieval from the PCRF, and quota validation from the OCS happening within each transaction.
The cumulative signalling load during a morning peak can generate millions of Diameter messages per second across the operator's core network interfaces (S6a, Gx, Gy, S11). Network equipment vendors design their platforms specifically to handle these burst loads — with Diameter Agent/Proxy nodes load-balancing and routing these messages across redundant server clusters.
Signalling Storm: What Operators Watch For
A signalling storm occurs when a large number of devices simultaneously re-attach to the network — often triggered by a brief service outage followed by restoration. When thousands of devices attempt to re-attach within seconds of each other, the MME signalling load can exceed design thresholds, potentially causing cascading delays. Operators mitigate this risk by implementing back-off timers in device firmware that randomise re-attach attempts across a time window, distributing the load.
Daily Traffic Patterns: The Network's Heartbeat
Mobile network traffic follows remarkably predictable daily patterns that operators use to plan capacity and schedule maintenance. A typical weekday traffic profile shows three distinct peaks: a morning peak (7–9 AM) driven by commuter attach events and morning browsing; a lunchtime peak (12–2 PM) driven by streaming and social media; and an evening peak (7–11 PM) that represents the heaviest data load, driven by home video streaming, social media, and gaming.
Overnight (1–6 AM) represents the network's lowest traffic period — typically 15–25% of peak hour volume. This is when the majority of planned maintenance activities are scheduled, software updates are pushed to network nodes, and capacity planning reports are generated from overnight data collection cycles.
These patterns have significant implications for how operators design and dimension their networks. Capacity planning is based on the busy hour traffic volume — the peak hour of the busiest day. Networks must be designed to handle this peak without degradation, meaning substantial capacity sits underutilised during off-peak hours.
Handover Event Volumes: A Network Never Stands Still
In an urban mobile network with millions of subscribers, handover events are among the highest-frequency operations the network performs. A single user travelling on a city train may trigger 20–50 handovers during a 30-minute journey — each one requiring coordination between base stations and in some cases the core network.
At the network level, this translates to an almost unimaginable event rate. A city-wide operator with 3 million active subscribers might process 50–200 million handover events per day across its RAN infrastructure. The vast majority — roughly 95% — are X2 handovers handled directly between neighbouring eNodeBs without involving the MME. The remaining 5% are S1 handovers (involving the MME, typically for inter-frequency or inter-RAT transitions) which carry higher signalling overhead.
The RAN Operations and Maintenance (OAM) system continuously collects Key Performance Indicators (KPIs) from base stations including Handover Success Rate (HOSR), Handover Preparation Failure Rate, and Handover Execution Failure Rate. A healthy network maintains an HOSR above 99.5%. A drop below this threshold triggers automated alerts in the Network Management System (NMS) and may initiate automated parameter optimisation processes or a manual engineering investigation.
Charging System Operations: Real-Time Accounting at Scale
The Online Charging System (OCS) is one of the most performance-critical components in the mobile network — and one of the least visible to users. It must process Diameter Credit Control messages from P-GWs across the entire network in real time, maintaining accurate quota counters for millions of prepaid subscribers simultaneously.
A typical OCS transaction involves: receiving a CCR-U (Update) message from a P-GW, querying the subscriber's quota record, updating the counter, computing a new quota grant, and returning a CCA-U (Answer) — all within a target response time of under 100 milliseconds. At scale, an operator-grade OCS may process hundreds of thousands of such transactions per second.
Throughout the day, the OCS records key events: plan activations and renewals at the start of billing cycles, quota threshold crossings that trigger subscriber notifications, exhaustion events that trigger policy changes, and recharge transactions that restore quota and reverse policy blocks. Each of these events generates a billing record archived in the operator's data warehouse for regulatory compliance and audit purposes.
From the OCS perspective, a recharge (top-up) event is a quota restoration transaction originating from the BSS. During peak hours — particularly evenings — recharge transactions represent a significant OCS event volume as users who have exhausted their daily or monthly data allowances seek to restore access. The OCS processes each recharge as a quota credit operation, followed by an immediate Diameter Re-Auth-Request (RAR) to the P-GW instructing it to re-evaluate the subscriber's policy — effectively lifting the data block within seconds.
Congestion Management: Dynamic Network Intelligence
Mobile networks employ several real-time mechanisms to manage congestion and ensure equitable resource sharing when cell capacity is under pressure. These mechanisms operate autonomously, adjusting parameters continuously without operator intervention.
Scheduling and Resource Block Allocation
The eNodeB's MAC scheduler dynamically allocates Physical Resource Blocks (PRBs) to active users every 1ms Transmission Time Interval (TTI). Scheduling algorithms — typically Proportional Fair — balance throughput maximisation against fairness, ensuring no single user monopolises capacity during congestion.
SON: Self-Organising Networks
Modern RAN deployments include Self-Organising Network (SON) capabilities — automated algorithms that continuously optimise coverage, capacity, and interference. SON functions include Automatic Neighbour Relations (ANR), Mobility Robustness Optimisation (MRO), and Capacity and Coverage Optimisation (CCO).
At the core network level, the PCRF plays a critical role in congestion management through Application Function (AF) signalling. When the network detects congestion conditions, it can push updated PCC Rules to the PCEF reducing the Maximum Bit Rate for non-priority subscribers, effectively sharing available capacity more equitably across all users in an affected cell.
Overnight Maintenance: The Network That Never Fully Sleeps
Despite lower traffic volumes, overnight periods are far from idle for mobile network infrastructure. The reduced load provides the maintenance window that operators use for essential operational activities that cannot be performed during peak hours without impacting service quality.
Software upgrades to network nodes — base stations, MMEs, P-GWs, OCS platforms — are typically scheduled in overnight maintenance windows. Modern network nodes support hitless upgrades using redundant hardware: one hardware unit is upgraded while the other continues handling traffic, then roles are swapped. This allows software to be updated without subscriber impact, but still requires careful coordination and is best performed at minimum traffic.
Overnight also sees the execution of batch billing processes in the OFCS: generating monthly invoices for postpaid subscribers, running data reconciliation between the OCS and BSS, archiving CDRs to long-term storage, and producing network performance reports for operational teams. The data generated by a large mobile network's charging systems in a single day can run to terabytes of CDR records.
Capacity planners also run overnight analytics on the day's RAN KPI data — identifying cells approaching congestion thresholds, monitoring spectrum utilisation trends, and flagging sites for capacity upgrades. This continuous, data-driven capacity management process is what allows operators to stay ahead of growing traffic demand and maintain acceptable quality of experience for subscribers.
The Scale of Daily Network Operations
To contextualise the engineering achievement that a modern mobile network represents: a large national operator may handle over 1 billion data sessions in a single day, process 500+ million OCS charging transactions, execute 100+ million handovers, and carry petabytes of data — all while maintaining 99.99%+ availability and sub-100ms charging response times. This is the scale at which mobile network systems operate every day, for every subscriber, invisibly and continuously.