NETWORK QoS ADVANTAGE: UTCP
UTCP is a novel congestion control algorithm
targeting both emerging wireless networks such as
LTE, WiMax, Wi-Fi, HSPA as well as high speed long
delay (high BDP) networks.
The Transmission Control Protocol (TCP) is a reliable
transport layer protocol that is widely used on the
Internet. It is the underlying protocol for HTTP,
RTMP and many other applications. Generally
speaking the congestion control algorithm is an
integral module of TCP that directly determines the
performance of the protocol. Standard congestion
control algorithms such as TCP-Reno and TCP-NewReno performed well for several decades but
are found to perform poorly over wireless and high
Bandwidth Delay Product (BDP) links.
To improve TCP performance over wireless and
high BDP networks, many TCP variants have been
proposed, including TCP Westwood, TCP Veno for
wireless applications and Compound TCP, TCP
CUBIC and FAST TCP for high BDP networks.
Although these algorithms have achieved
performance increases in their respective target
applications, designing a TCP congestion control
algorithm that performs equally well in both
wireless and high BDP networks is still a challenge.
On the other hand, with the deployment of wireless
networks such as LTE or WiMAX and high-bandwidth,
real-time applications such as multimedia over
TCP/HTTP, the TCP congestion algorithm is required
to handle both wireless connections with radio-related losses as well as congestion-introduced
issues typical for wired high BDP networks.
The UTCP congestion control algorithm achieves
major improvement in this highly important
area. The quality increase in stream viewing and
broadcasting gained by improving this technology
puts Ustream ahead of alternatives in terms of
streaming QOS and delivery costs.
The performance of UTCP on a Wireless network
with the following attributes: 5Mbps available
bandwidth, 1% packet loss, 100ms round trip time.
QUALITY OPTIMIZATION OF DELIVERY
Ustream utilizes several delivery sources, some of
which are internal while others are global 3rd party
CDNs or regional 3rd party CDNs.
The choice of delivery source is optimized on a
per-user basis, prioritizing for optimal quality. To
execute this, Ustream collects real-time usage
information about the peering lines and transit
lines as well as usage on the supported CDNs
and also geo-based quality information from
The simplified decision-making flow is as follows:
• Select an edge cluster based on the geolocation
of the visitor.
• Based on the user’s IP address and AS code
check if Ustream can serve the content from
any of the peering lines supported in the
given edge cluster. Use a peering line with free
capacity if available.
• In the absence of peering capacity check if there
are any usable transit lines with free capacity
on the given edge cluster. There is an algorithm
managing usage on transit lines and optimizing
them for balanced 95th percentile traffic.
• When falling back to a third party CDN, CDN
choices are prioritized partly by historical
performance in the given region.
Traditional TCP congestion control algorithms are
found to perform poorly over wireless networks:
• Traditional TCP: Packet loss = Congestion
• Wireless networks: Packet loss ≠ Congestion
Sponsored Content IBM Cloud Video