DOZER - Technical Overview
DOZER - Technical Overview
DOZER - Technical Overview
Table of Contents
Executive Summary
1.2
2.1
2.2
2.3
3.1
3.2
3.3
3.4
3.5
3.6
10
4.1
4.2
4.3
Point-to-Point ............................................................................................................................. 10
4.4
4.5
4.6
Conclusion .................................................................................................................................. 11
4.7
12
Executive Summary
Video operators need to transport high-quality and time critical video streams at the lowest possible cost,
which increasingly leads them to explore IP network based approaches including the public Internet. However,
during transmission across an IP network a packet may be delayed with no guarantee of arrival at a specific
time, or in the same order as transmitted. Since video is extremely time sensitive it follows that transmission of
high-quality real-time video over IP networks is very challenging.
The light weight UDP real-time data delivery protocol was designed for streaming video applications but
assumes network nodes are not congested. However, a key problem is that IP network routers will simply drop
UDP packets when congested, without notifying the sender. This is a critical issue with video delivery over IP
networks, especially when UDP video has to compete with TCP/IP data traffic. It is well understood that
streaming video using UDP is affected by TCP traffic and its demanding flow control mechanism whenever TCP
and UDP share a node, and hence the UDP transmission may experience packet loss.
TCP, a heavy weight protocol by comparison, is neither suitable for live video distribution, nor high bitrate live
ingest to Content Delivery Networks (CDN). This is due to the protocol's flow and congestion control
mechanism, which interprets packet loss as evidence of bandwidth limitations and in response reduces the
packet transmission rate. This is exacerbated over longer distances, and when traversing multiple networks,
both which tend to increase the packet loss rate, thus reducing the effective throughput.
HTTP-based adaptive bitrate (ABR) protocols are designed to ensure the best possible video quality for any
given bandwidth and screen resolution. The client device selects the video stream that is most optimal to local
circumstances, but that is by itself not a guarantee of a broadcast Quality of Experience (QoE).
Therefore, to maintain the quality of real-time video transmitted over IP networks, while also minimizing
transport cost by utilizing the Internet, a different approach is required.
Enter the award winning "DOZER" technology, integrated in DVEO's encoders/transcoders and receivers.
DOZER literally "bulldozes" real-time video through IP networks, including the Internet, in a deterministic
manner thanks to its ability to automatically correct for UDP packet loss, thus safeguarding the video quality.
DOZER is a real-time IP Gateway technology featuring an end-to-end
error correcting protocol that delivers time critical video reliably over
UDP, whether unicast or multicast. It enables smooth MPEG-2 and H.264
SPTS and MPTS delivery of both SD and HD services through DVEO
patent pending Automated Packet Recovery algorithms for alleviation of
packet delay variation and outright packet loss, and re-ordering of outof-sequence packets.
The DOZER is a cost-efficient solution for a variety of UDP video applications by making
them virtually immune to out-of-sequence packets and outright packet loss, and using a
minimal sub-second latency buffer necessary for accurate, high quality video transport.
The DOZER is commercially available in three versions enabling error-free UDP video transport for contribution
or distribution over all types of IP networks including the public Internet:
1. DOZERbox II IP/IP, a compact stand-alone product (pictured).
2. DOZER Racks IP/IP, 1 RU.
3. DOZER APR: LIC, a software license for integration in third-party products, and an available option in
DVEO encoders/transcoders and receivers. This version allows customers to originate and terminate
video delivery without requiring additional hardware devices, thus reducing both CAPEX and OPEX.
June 15, 2015
1.1
1.2
2.1
TCP-based protocols
are neither suitable for
live video distribution,
nor high bitrate CDN
ingest.
However, since the packet recovery time will vary from one congestion
scenario to another, TCP-based protocols are neither suitable for live video
distribution, nor high bitrate live ingest to CDNs. This is due to the protocol's
flow and congestion control mechanism, which interprets packet loss as
evidence of bandwidth limitations and in response reduces the packet
transmission rate. This is exacerbated over longer distances, and when
traversing multiple networks, both which may increase packet loss rates, thus
reducing the effective throughput.
Another protocol such as the Real Time Protocol (RTP), which runs on top of UDP, supports packet re-ordering
and de-jittering, to a degree, but like UDP it does not offer packet loss recovery.
While there are some potential solutions to these issues, such as Forward Error Correction (FEC) protocols,
they either require substantial additional bandwidth permanently and/or will delay real-time video delivery in
an undesirable manner.
A potential alleviator of network bandwidth variations is Adaptive Bitrate Streaming (ABR), with protocols such
as Apples HTTP Live Streaming (HLS), Microsofts Smooth Streaming, Adobes HTTP Dynamic Streaming (HDS)
and the emerging MPEG-DASH standard (Dynamic Adaptive Streaming over HTTP). Common to all is that they
will encode the video at various bitrates, ranging from HD quality to lower bitrates that may offer a basic or
even decent experience but not a broadcast QoE. The client device selects the video stream that is most
optimal to local circumstances, which ensures the best possible QoE for any given bandwidth and screen
resolution. This is useful for multiscreen video services to devices of varying display capabilities and across
various network types, but that is by itself not a guarantee of a broadcast QoE where it matters most.
While CDNs are increasingly used to overcome real-time video quality issues,
this is limited to the traffic within the CDN infrastructure. So, no matter how
solid a CDN is internally, if the video quality is impacted already between the
video origination point and CDN ingest, no CDN can restore it. Therefore,
special care need to be taken when transmitting live video from its source to
the CDN ingest facility.
2.2
Packet Delay Variation: Minor variations in packet arrival time, sometimes referred to as jitter, occurs
due to network congestion, timing drift, or route changes. Variations in packet latency across a
network can cause problems if the latency exceeds the jitter buffer size as defined in HW or SW.
Packet Loss: UDP packets may not arrive at all, since UDP does not have a built-in packet recovery
mechanism, and with no notification provided to the sender. The probability of packet loss increases
linearly with distance since the number of hops traversed is generally larger. When and if this occurs,
video quality is impacted and at some point it becomes unacceptable.
Dedicated and over-provisioned network infrastructure has long been the norm for telcos and cable operators,
as well as dedicated CDNs, to minimize jitter and packet loss rates. Such networks offer near perfect Quality of
Service (QoS) but at a price and, nevertheless, are still subject to discontinuities and (albeit low) packet loss.
This is particularly egregious when video traffic is routed over "bursty" fiber connections.
Video over Local Area Networks (LAN) generally offers a near perfect QoE. Provided there is sufficient
bandwidth and little TCP traffic, congestion can be really low although rarely eliminated. It turns out that
switches and routers are not yet perfectly tuned for video and quite often introduce jitter, discontinuities and
dropped packets. This is caused by bandwidth contention between TCP and UDP packets.
2.3
3.1
3.2
DOZER Applications
The DOZER is a cost-efficient solution for a variety of UDP video applications by making
them virtually immune to out-of-sequence packets and outright packet loss, and using a
minimal sub-second latency buffer necessary for accurate, high quality video transport.
For example, customers in the U.S. use DOZER technology to import video content from countries such as
Bangladesh, Colombia, Greece and Iraq with no packet loss or video freezes, and lower latency than FEC-type
protocols. An Asian IPTV operator uses the DOZERbox to reliably transfer video between cities, and a TV
station in American Samoa between islands. Customers also create low cost CDNs over the Internet. In other
words, DOZER technology improves video delivery over both the Internet as well as managed networks.
3.3
3.4
3.5
DOZER Versions
The DOZER is available in three versions enabling error-free UDP video transport for contribution or
distribution over all types of IP networks including the public Internet:
1. DOZERbox II IP/IP, a compact (4.6 in. / 117 mm square) stand-alone product as pictured in Figure 1.
2. DOZER Racks IP/IP, 1 RU form factor for rack mounting.
3. DOZER APR: LIC, a software license for integration in third-party products, and an available option in
DVEO encoders/transcoders and receivers. This version allows operators to originate and terminate
video delivery without requiring additional hardware devices, thus reducing both CAPEX and OPEX.
DOZER is an available option in a wide range of DVEO products, such as the MultiStreamer and
Mamba DIG/IP encoders/streamers, Gearbox and Gearbox II transcoders, Atlas Media Server and
Brutus transcoder/media server.
3.6
4.1
Application Versatility
Studies and real world usage indicate that each type of network, whether LAN, WAN or dedicated and
conditioned network, can be perfected by a properly configured pair of DOZERS or DOZERIZED encoders,
decoders and gateways. When video packets transits corporate networks, packet loss is likely low. When
packets transit dedicated lines, packet loss is higher. When video packets transit public Internet then packet
loss is obviously much higher.
The DOZER is a cost-efficient solution for broadcasters, mobile operators and telcos for a variety of application
and deployment scenarios by making transmissions virtually immune to variable latency, out-of-sequence
packets and outright pack loss, and by not delaying time critical video like FEC technologies do.
DOZER ideal for:
Delivery of multiscreen services over networks that do not offer QoS guarantees such as the Internet
Multi-network and multi-country transfers for content producers and service providers
Streaming studio quality live news and sports programming over congested backbone networks
Replacement of expensive point-to-point managed network connections and traditional satellite
backhaul by creating low-cost CDNs using any type of network including the public Internet
Enhance video delivery reliability and quality also over managed networks where UDP video
transmissions have to compete with TCP/IP data traffic the latter is prioritized by network routers
and will lead to dropping of UDP packets when the router buffers fill up. The DOZER technology will
recover the lost packets with extremely low latency, thus preserving the real-time video delivery.
4.2
Multi-Island Distribution
American Samoa KVZK-TV was the very first television station in the South Pacific,
celebrating its 50th anniversary in 2014. It was facing a major challenge in distributing its
content over IP to the five islands of the territory. They tested and eventually chose IP video
encoding equipment from DVEO, also outfitted with the DOZER technology. This solved their
content distribution problem and saved money per Daniel Langkilde, Director of the Office of
Public Information at KVZK-TV: "Our American Samoa TV network was having major interisland communications issues with IPTV. The DVEO encoders with built in DOZER saved us
large amounts of money by not having to use Satellite Links between islands."
4.3
Point-to-Point
One of the top three U.S. broadcasters uses DOZER to connect between East and West Coast facilities. They
found noticeable packet loss on their dedicated lines and now rely on DOZER on top of their Telco conditioned
lines. The video drop outs are now non-existent.
4.4
Multi-Country Distribution
Ellas TV is an ethnic channel content originator. It needed to bring content from Greece to the U.S. and found
the DOZER to be extremely price effective when compared to using satellite, dedicated lines, or CDNs.
4.5
10
4.6
Conclusion
As can be seen from the above use cases, DOZER technology benefits UDP video transmissions over any type of
IP network, whether the public Internet or dedicated lines. It alleviates packet loss and various other forms of
delivery problems such as packets arriving out of sequence.
4.7
Application Note: Using Linktropy WAN Emulation to Demonstrate DOZER Automated Packet Recovery
11
12