The technical challenge of measuring the flow is to know at which point the traffic is collected, whether it is up and down, how P2P flows are recognized, how the outstanding traffic after user breaks is handled and how excess reminders are sent. If these problems are not resolved, counting the flow charges is a mess.
Collection points for traffic statistics
The flow charge is first to determine which equipment the traffic is. Three collection points are available: AC statistics, Authentication Gateway statistics, and independent flow analysis devices. The advantage of AC statistics is that no additional equipment is needed, but AC traffic statistics are not accurate, usually only by type of application, without distinction between the IP layer and the AC performance will decline due to a large number of user flows. NATSHEL_AUTH_GATEWAY (e.g. NatShell NE-80) has a higher level of accuracy, supports the DPI in-depth package testing and identifies more than 3600 applications, and the gateway itself is on flow routes with no additional delay. The independent traffic analysis device collects traffic via mirrors with the advantage that it does not affect network performance, the disadvantage being that mirror flows may be discarded and statistical precision restricted by mirror bandwidth. The campus recommends best balance of accuracy and performance on flow billing at Authentication Gateway.
You're counting up and down?
The flow-by-flow charge is either down (download) or up/down row. Most campuses are down by traffic, because the main traffic consumption of users is downloading (seeing videos, browsing web pages and downloading files), usually with minimal upward traffic (message, upload pictures). But some scenes require up/down-and-down, such as video monitoring uploads in the park, cloud backup, etc. The billing rules are clearly defined in the user agreement to avoid complaints when users find that they have been charged for going up or down.
Special handling of P2P traffic
P2P downloads (BT, ass, and thunder) are the most painful problem in flow-based charges. The P2P traffic is characterized by large up-to-down rows and connections, with one user using P2P downloads to fill the full export bandwidth. Under flow-related costs, P2P users consume dozens of times more than ordinary users, at a prohibitive cost, and see post-billing complaints from users. There are several ways to handle them: the first is that the P2P traffic is either uncosted or priced separately, and less expensive than normal flows, encouraging users to use it during non-peak periods. The second is the P2P traffic limit, limiting the maximum bandwidth of single-user P2Ps and avoiding affecting other users. The third is a P2P traffic account for total flows but with a daily ceiling that exceeds the post-P2P speed.
Flow settlement after user disconnect
When the user normally goes down, the system sends a stop message to Radius' billing, which includes the total traffic on the current line and a cut-off charge. But users may break their wires abnormally: handheld machines, WiFi signal loss, AP malfunctions, network interruptions, where there is no normal downlink? The V7 system is handled through two mechanisms: the first is real time, when the user sends an intermediate report every time (such as 15 minutes), when the user is online, when the traffic is reduced in real time, and the traffic that was deducted at the time of the anomaly is actually used, and the last intermedian to the break-line is negligible because it is small. The second is idle time, with no traffic for a period of time (e.g. 15 minutes), and the system automatically considers the user to be offline and the billing costs are stopped.
The traffic overflow alert.
The traffic charge must be over-alerted, and users complain that no advance notice is provided when the user has finished using the flow. Reminder is divided into several nodes: one reminder at 50 per cent, to inform users of the progress in use; a second reminder at 80 per cent, to recommend upgrades or savings for users; a third reminder at 100 per cent, to advise that the traffic is out and will be downped or down. A reminder is used as text messages because the user may not be able to access the Internet and send them. V7 supports SMS notification functions, which can configure the flow threshold alert template and automatically detects the user traffic usage and sends text messages when reaching the threshold. Reminders are not too frequent, users will feel harassed and 50 per cent, 80 per cent and 100 per cent of the three nodes are sufficient.
Treatment after exhaustion of traffic
The user traffic is treated in two ways: shut down or slow down. Cutting off is a direct break of the network, and users have to buy a flow pack or upgrade a package before they can recover. The deceleration is continuous but the bandwidth is low (e.g. 128Kbps), so users can send messages but cannot see videos and guide users to purchase traffic. The experience with the slowdown is better than when the shutdown is done, the user will not completely disconnect the net, but the bandwidth after the drop does not need to meet basic communications needs and cannot be reduced to even micromails.
Reconciliation of traffic statistics and dispute resolution
The V7 system records the down-to-down time for each user, traffic consumption per line, and application type distribution. The user can search its own flow details on a self-service page. Dispute processing process: The user objects, the administrator asks the user ' s traffic log to confirm that the statistics are accurate. If there is a system statistical error (e.g. an abnormal breakout leads to extra counting), refunding the overcharges and recording them. If it is the flow that the user itself uses (e.g. automatic backstage updates), show the user a clear explanation. The accuracy of traffic statistics is regularly verified, the SMS statistical and physical values are compared using standard flow testing tools, with errors being kept within 5 per cent.
The campus WiFi Authentication & Billing system collects the most accurate and timely traffic charges on NATSHEL_AUTH_GATEWAY, with clear rules for reporting up-to-down, separate P2P traffic, real time charge extra hours for an anomaly line, 50 per cent/80/100 per cent reminder that flow landings are better than stoppages after exhaustion, and that the details of traffic can be checked.