Video on Demand (VOD) is a service that delivers the video content you want in real time, and it uses different delivery methods, such as RVOD and NVOD. RVOD provides greater user convenience but is limited by bandwidth, and NVOD controls latency but can be less efficient with network resources. Each method impacts user experience and network efficiency.
Video on Demand (VOD) refers to a service in which a server transmits video content over the network in real time in response to a user’s request, and simultaneously plays it back on the receiving end. These services differ from traditional broadcast systems in that they allow users to access the content they want at any time. The flexibility of VOD has made it the preferred method for many users, and this has led to the development of many different service models.
When content is delivered in real time, it’s important that the data is delivered within an acceptable timeframe, so it takes the form of spreading the data over the network, much like an over-the-air broadcast. This is why real-time streaming is important, as users need to be able to watch video content without interruption. The delivery of content occurs through software-defined channels, which help data flow smoothly like lanes on a highway. A channel acts as an outlet for blocks of content data, and its size is represented by a “band,” which is the amount of data it transmits per unit of time. The more data that is transmitted, the larger the bandwidth required, which is essential for delivering high-quality video.
On the other hand, the size of the band that a server can accommodate, or its maximum transmission capacity, is called “bandwidth” and is expressed in bits per second. The higher the bandwidth, the more users can be served with high-quality video at the same time. However, bandwidth is not infinite, and service providers need to manage this resource efficiently. Advances in network infrastructure are essential for this, and next-generation network technologies such as 5G are expected to play an important role in improving the quality of VOD services.
Among the various methods of VOD, the method that creates a separate channel for each user’s request is called RVOD (Real VOD). Since each transmission channel is independent of each user, users can control real-time transmission such as ‘pause’, ‘fast forward’, etc. directly, which is relatively convenient for users and enables simultaneous service of various contents even with limited bandwidth. The biggest advantage of RVOD is the ability to optimize the user experience. Users can freely manipulate the content they want at the moment they want, which is the key to providing personalized services. However, the overall amount of data that the server has to transmit increases in proportion to the number of concurrent users, so there is a limit to how many users can connect at the same time if bandwidth is limited. This has the potential to degrade the quality of service during peak hours, which requires a technical solution to overcome.
Near VOD (NVOD), which is proposed to overcome this drawback, is a method of bundling service requests received during a certain period of time and providing service through a form in which multiple recipients are simultaneously connected to one channel. Since one channel of NVOD requires a certain bandwidth regardless of the number of simultaneous subscribers, it can overcome the limitation of the number of simultaneous users, but it is inconvenient that users have to wait for a certain period of time to receive the service. This is similar to the concept of traditional broadcasting, where the broadcast time is set and viewers must tune in at that time. From a service provider’s perspective, the amount of waiting time a user can tolerate without canceling a service request is called “tolerable latency,” and it is an important factor in determining the quality of VOD. Minimizing the tolerable latency is one of the key factors that determine the success of NVOD, and an efficient scheduling algorithm is essential for this.
“Time-sliced NVOD” is a way to reduce latency by having the same content repeatedly delivered across multiple channels at staggered intervals. The goal is to minimize the wait by having the user listen on the channel with the lowest latency after the time of the request. The user listens on the channel with the lowest latency after the time of the request and starts listening when the transmission on that channel resumes from the beginning of the data block. The latency is determined by the number of channels on the server and the length of the content: if a 120-minute movie is delivered over 12 channels at 10 minute intervals, the latency will be less than 10 minutes. To reduce the latency, a large number of channels are needed, such as 120 channels to make it less than 1 minute. In this approach, latency is directly affected by the number of channels, so the cost of the service provider’s infrastructure becomes an important variable.
Data-split NVOD controls latency by dividing content into multiple data blocks and delivering each separately on multiple channels. This method is designed to increase delivery efficiency and offers a different approach to traditional time-slicing. By making the first block a reasonable size, the entire content is divided so that the number of blocks is as large as the number of available channels, with subsequent blocks sequentially doubling in size. Each channel starts sending its allocated blocks in sequence at the same time, and repeats the transmission periodically based on the size of each block. The receiving end receives the content in order from the first block after the time of the request, and when it finishes receiving a block, it automatically switches to the channel where the next block is being transmitted and receives it from the beginning of the block. However, the bandwidth of the channel must be at least twice as large as that required to play the content, so that the reception of the next block is guaranteed while the already received portion is playing, and continuous playback is possible.
This method reduces latency because the size of the first block is relatively small. In the previous example, if 120 minutes of content is served over 6 channels at 2x speed, the latency is less than 1 minute. This can be an important factor in user satisfaction, and in fact, this approach may be more positively received by latency-sensitive users. Thus, compared to time-slicing, latency can be reduced by more than 90% while occupying the same amount of bandwidth, and the number of channels used relative to latency can be reduced to increase the amount of content that can be served simultaneously on a single server. However, since the transmission time of the entire content is less than half of the total playback time of the content, each channel must maintain more than twice the transmission capacity, and the receiving end must have enough space to store data for half of the content.
NVOD commonly utilizes multiple channels for latency control, so there is always a certain amount of bandwidth available for the service. This is an important consideration when designing an NVOD system, and the scalability and reliability of the network infrastructure is key. Therefore, it is not very efficient when there are few concurrent users per content. Even in the extreme case of a single user, the bandwidth required for six channels in the above example would be occupied, resulting in a significant waste of network resources. To compensate for this, it may be necessary to select different NVOD methods depending on the nature of the content, or to introduce a system that automatically selects the most appropriate delivery method by analyzing user patterns. Such a system can efficiently utilize network resources, allowing more users to access the service simultaneously.
