Dynamic/Adaptive QoS

Many Wi-Fi gateway or controller solutions can control user bandwidth, but the setting is usually a static fixed configuration. For example, if the configuration sets 2Mbps per user, each user will be capped at 2Mbps at all time, regardless of the overall usage of the ISP link. For a typical public Wi-Fi, each user is limited to maximum speed of 2Mbps, or 5Mbps, etc.

However, statically setting per user bandwidth is unable to efficiently utilize available ISP link capacity and unable to maximize user experience. For example, when the no. of concurrent users is low, each user is still capped at pre-configured speed (eg. 2Mbps), and they are unable to burst effectively to use the spare capacity. Particularly in countries like Singapore, where the fiber broadband is widely deployed with huge bandwidth at each location, it’s very common for a hotpot network to have 500Mbps link. When there are total of 10 active users, if each user is capped at 2Mbps, they can only use total of 20Mbps, wasting all other available capacity. This causes a lot of wastage on the available network resource, while the user experience is not maximized.

RansNet HotSpot Gateway (HSG) and HSA have some advanced QoS features to dynamically or adaptively allocate per user bandwidth by correlating the backhaul usage with concurrent user sessions, to maximize user experience and ensure fair use of the network resources.

Dynamic Bandwidth Management (DBM)

With dynamic bandwidth management capability, instead of statically fixing a per user bandwidth, HSG/HSA correlates the subscribed backhaul bandwidth and actual usage (no. of active devices) dynamically allocate per user access speed, so that we can fairly and fully utilize the available backhaul bandwidth.

E.g.: Assume a subscribed backhaul of 100 Mbps,

No. Active Device Max Access Speed Allocated to Active Device-------------------------------------------------------------------------------------------1 2 Mbps or 100 / (0 + 1) = 100 Mbps whichever is higher2 2 Mbps or 100 / (1 + 1) = 50 Mbps whichever is higher..........n 2 Mbps or 100 / (n + 1) Mbps whichever is higher

In order for this dynamic access speed allocation to work accurately, there are two key inputs:

    • backhaul bandwidth. Backhaul speed is the numerator in above equation, and its value must be equivalent to the upstream link speed in order for theoretical results to be consistent with actual experiences.

    • no. of “Active” devices. This is the denominator in above equation, and it’s very important to be accurate. Particularly in the case of open public Wi-Fi network, many phones/devices are auto associated to the Wi-Fi SSID without users’ awareness (without users actively using the network). If we count all the associated devices as “Active” devices, the total device nos. will be falsely high, and allocated per device access speed will be falsely low, ending up with inefficient use of backhaul again. Our HSG only counts the authenticated users as “active” users.

Configuration of dynamic bandwidth allocation can be done on a "global" or per instance (per SSID) basis. Each instance is mapped to a VLAN or network, with the allocated upstream backhaul bandwidth (PS: below command is applicable to both HSG and HSA).

!security hotstpot vlan10 client-bandwidth dynamic <backhaul_download> <backhaul_upload> minimum <min_download> <min_upload> maximum <max_download> <max_upload> .................!
    • <min_download> <min_upload> is the min guaranteed per user bandwidth. However, do note that if the backhaul is extremely congested, no user speed can be guaranteed. eg. for a 100Mbps link, if we configured min 2Mbps per user, but if there’re more than 100 active users, and each user is all actively browsing/downloading, they may not be able to have guaranteed bandwidth.

    • <max_download> <max_upload> is the maximum capped bandwidth during burst. This is an optional entry. Sometimes it’s desirable to cap per user bandwidth even when the usage is low. For example, if there’s only 1 user with a 100Mbps backhaul, without maximum cap the user will be dynamically allocated with 100Mbps bandwidth, which can be a potential risk, eg. one user could use the local connection (with huge backhaul) to launch DDoS attack to external networks.

Adaptive User Bandwidth Control (AUBC)

AUBC is a RansNet patented bandwidth control technology to manage per user QoS more efficiently. It is a more advanced technique targeting at public hotspot networks or visitor based networks, such as public Wi-Fi network, eg. shopping malls, cafes, retail outlets, and hospitalities network eg. hotels, hostels, hospitals, etc.

HSG/HSA correlates the actual ISP link usage with its subscribed capacity and adaptively adjusts per user speed, so that each user can burst to an allowable speed/bandwidth when the Internet link usage is low, and slow down when the Internet link is reaching congestion.

This technique will fully utilize ISP link capacity to maximize user experience. There are two techniques used in the AUBC process:

  • Real-time monitoring of ISP link utilization

  • Controlling per user bandwidth based on proprietary algorithms

AUBC can be "global" or specific to a hotspot instance only:

  1. when it's applied globally, HSG/HSA tracks the actual link usage and applies the bandwidth control to all hotspot users (across all VLANs/instances)

  2. when it's applied to a particular VLAN/instance only, mbox tracks the bandwidth usage for the instance only and applies the bandwidth control to users within the instance.

Below command apply to global control:

!security hotspot adaptive eth0 10000000 10000000 1 maximum 5000000 5000000 minimum 2048000 2048000!

where:

  • 10000000 is the total backhaul (ISP link) upload/download speed

  • 1 is the tracking interval (in minute)

  • 5000000 is the maximum per user speed (upload/download)

  • 2048000 is the minimum per user speed (upload/download). Note this is not guaranteed if the back-haul is congested.