Video communication over wireless channel has experienced extensive growth in the last two decades. Wireless networks are expected to provide Quality of Service (QoS) that guarantees end-to-end delay and end-to- end distortion. To this end, many works have been conducted to analyze the complexity, rate, and distortion performance of the hybrid video encoders. It is also applied in the wireless video communication system to reduce the end-to-end distortion, and also the optimized rate control scheme was developed (Li, Xiong, & Wu, 2015). This rate control technique attempts to reduce both video encoding distortion and transmission distortion (due to transmission rate and delay).

A block of an end-to-end wireless video communication system is shown in Figure 1, which consists of five parts, i) the video encoder, ii) encoder buffer, iii) the video decoder, iv) decoder buffer, and v) the error control channel, which is further divided into channel encoder, wireless channel, and channel decoder (Li et al., 2015; Stuhlmuller et al., 2000). End-to-end distortion can be minimized under delay constraints (Hsu, Ortega, & Reibman, 1997). According to Li et al. (2015), rate can be optimized by considering the end-to-end delay bound, by combining both source coding and channel coding parameters.

Figure 1. Block Diagram of an End-to-End Wireless Video Communication System

To achieve optimal performance under the delay and energy constraints, by using the traditional R-D model and the previously proposed d-R-D model, a novel delay-power- rate-distortion (d-p-r-d) model (Li, Wu, & Xiong, 2014) was formed including another two dimensions (the encoding time and encoder power consumption), which quantifies the relationship among source encoding delay, rate, distortion, and power consumption.

In video communication system, the video which is to be transmitted is encoded then converted into packets of variable length and then finally multiplexed with audio. In between the source and destination, a dedicated link exists that can provide a guaranteed QoS. Packets or data bits may be corrupted or vanished due to traffic congestion or bit errors due to impairments of the physical channels. With advanced radio networks, the current Internet makes it hard to provide QOS, such as, packet loss probability, rate, bandwidth, and delay needed by video communication applications. Higher video quality over wireless channels is difficult because wireless channels are time-varying and error vulnerable, which leads to lower throughput and higher bit error rate. Particularly for the process of capturing the picture to video display, the End to End delay experienced by each frame is composed of various delay components as shown in Figure 1. To evaluate the End to End delay, all these delay components must be carefully designed because change in one component would change the delay budget. So capture-to-display delay rate, distortion and power constraints should be considered for an efficient wireless video communication system performance.

Delay Standard: If the delay level is over 400 ms (milli seconds), the network standard is regarded as below standard (Table 1).

Table 1. Delay Standard

Many researchers have worked on rate control schemes (Hsu et al., 1999; Pudlewsti et al., 2015; Sühring, 2012) and few others are working on improving the video transmission to guarantee the QOS parameters via various approaches. Few of them have formulated the end to end distortion in terms of bit rate and packet error rate (Stuhlmuller et al., 2000; He, Cai, & Chen, 2002; Chen & Wu, 2012). Some works have been done on operational rate distortion functions (Kondi, Ishtiaq, & Katsaggelos, 2002). Compared with the operational models, the analytical models are more desirable because of their lower complexity (Stuhlmuller et al., 2000; He et al., 2002; Chen & Wu, 2012). These analytical models give a complete analysis of a video transmission system, such as For ward Error Correction (FEC), Rate-Distortion performance of the video encoder, the effect of error concealment, and inter-frame error propagation at the video decoder. The End to End wireless video communication was also analyzed by a cross layer optimization as in (Pudlewski et al., 2015).

Some of the recent contributions in this area are classified as follows

1) Delay Rate Distortion Optimization Rate (DRDOR)

2) Power supply constraints

3) End-to-End delay

• DRDOR: This was proposed by Li et al. (2015). This scheme developed a theoretical framework to optimize the rate, by proposing the analytical solution for joint source rate control under the time varying wireless channel conditions. It is designed to satisfy the transmission rate by considering the end-to-end delay and distortion. This technique is used to extend the rate distortion model for video encoder by using the statistical analysis of encoding delay (Chen, 2011; Chen & Wu, 2015).
• Power supply constraints: By using d-p-r-d model (Li et al., 2014) with four dimensions (delay-power-ratedistortion), source rate control was done which reduces the source coding distortion under encoding power, the maximum encoding delay, and source encoding rate.
• End-to-End delay: The analytical and simulation results demonstrated that end-to-end delay has significant impact on throughput and packet error rate (Gong & Wang, 2014). All the aforementioned works assume video encoding time T_e and decoding time T_d as constants and can be neglected. Therefore, the study of system performance under end to end delay is actually the study of system performance under queuing delay and transmission delay (Guan, Melodia, & Yuan, 2013). Besides the above issues, different video coding system structure leads to different rate control frameworks.

By all these works, the minimal video quality to all users was provided. But it is desirable to provide maximal HD video quality with guaranteed QoS to all the users.

To obtain the optimal QoS performance for the entire end-to- end wireless video communication system as shown in Figure 1, however, not only the source coding distortion needs to be minimized but also the transmission distortion should be considered (Chen & Wu, 2010). In addition to the constraint of maximum encoding delay, end-to-end delay constraint should also be satisfied, which requires the suitable allocation among the encoding delay and other delay components in accordance with the end-toend delay bound. Existing works on cross layer measures performance in terms of end-to-end distortion, packet loss rate, and bit rate. As encoding time has an effective impact on encoding rate (R) and distortion (D), the optimal R-D performance based on the assumption of infinite encoding time is neither accurate nor realistic. To achieve the optimal end-to-end performance under low delay constraint, the traditional R-D model is extended to form a novel delay rate-distortion (d-R-D) model, which quantifies the relationship among source coding time, rate, and distortion.

In real-time wireless video communication, the system has a maximum tolerable end-to-end delay for each video frame. The end-to-end delay consists of video encoding delay T_e, queuing delay at a transmitter buffer T_eb, wireless transmission/propagation delay T_c, queuing  delay at a decoder buffer T_db, and video decoding delay T_d (Hsu et al., 1999; Chen, 2011) as shown Figure 2.

Figure 2. Block Diagram of End-to-End Delay of a Wireless Communication System

To achieve the best end-to-end QoS performance in wireless video communication system, one needs to think out of the boundary of the subsystems and formulate the problem on QOS control for wireless video communication. The methods to minimize transmission distortion (Chen & Wu, 2010) under packet transmission delay constraint is researched, but transmission delay is often equated to end to end delay without further interpretation.

2.1 End-to-End Delay

The end-to-end delay experienced by a video frame is equal to a constant C, then

(1)

If we assume the constant video frame rate that is same at both encoder and decoder, the end-to-end delay per frame is required to be less than a maximum acceptable delay interval T_max, which is the end-to-end delay.

(2)

Assume d_c is constant, and d_d is neglected when compared with d_e. The allocation and trade-off between d_e and d_buffer will be given by equation (3).

(3)

 

The time spent by each frame to stay in the encoder buffer and the decoder buffer is given as follows.

(4)

To ensure real-time video encoding without introducing accumulated encoding delay for each frame, the maximum video encoding time at the encoder should not exceed the time duration of a frame interval.

(5)

where T_e be the encoding time, T_f be the time duration of a frame interval.

2.2 End-to-End Distortion

The end-to-end distortion D_ete is calculated as,

(6)

where D_S denotes the source coding distortion caused by quantization error during lossy video compression and D_T is the transmission distortion caused by transmission error due to bandwidth fluctuation and packet losses.

With the same channel characteristics in a wireless channel, the reproduced picture at the decoder will not be the same as it was at the encoder due to random packet losses. The tradeoff among delay, rate, and distortion for the design of the video communication system could be applied to control QoS.

2.3 Joint Source Channel Rate Control (JSCRC)

The Joint Source Control Rate Control algorithm can improve the system performance by using Markov channel model, Lagrange multiplier approaches, and Karush-Kuhn-Tucker (KKT) conditions. There are four major contributions in this work. First, traditional Rate-Distortion model determines the rate and distortion for any given intra refreshing rate. Second, along with Rate-Distortion functions, energy supply constraints are also included. Statistical channel-distortion model reveals the inherent relationship between the channel distortion and input video characteristics. It also calculates the channel distortion with minimum delay. Third, based on the source and channel-distortion models, a scheme for adaptive intra-mode selection and joint source-channel rate control was developed. Many experimental results have demonstrated that it considerably improves the video quality for wireless video communication systems. Finally, a DRD optimized rate control algorithm was developed to minimize the end-to-end distortion subject to the transmission rate and delay constraints (Li et al., 2015; Chen & Wu, 2010). This model has less computational complexity, so it can be applied for wireless video applications.

To achieve the best end-to-end QoS performance in wireless video communication system that targets at minimizing the average total end-to-end delay and endto- end distortion (Guan et al., 2013; Hsu et al., 1997), we have to consider two tradeoffs regarding source coding delay versus buffering delay and available source coding rate versus redundant rate incurred by channel coding were coupled in the optimization problem. Theoretically, a large search range as well as a small quantization step size is required to result in smaller source coding distortion with large source coding bit rate. However, a source coding delay is also enlarged while the number of frames stored in encoder and decoder buffers is decreased which in turn limits the available source coding bit rate supported by transmission channel and thus affects the source coding distortion.

In most cases, decoding time is too small with regard to queuing delay and can be neglected in end-to-end delay. However, video encoding time depends on video encoding complexity that can affect source distortion and bit rate as verified in the complexity rate distortion model of video encoding. Hence encoding time should be properly included in R-D performance and becomes Delay-Rate-Distortion (D-R-D) model. On the other hand, a constant end-to-end delay is given, if encoding time T_e increases, the tolerable queuing delay and transmission time will shorten, this in turn reduces the channel throughput, thereby increasing the transmission distortion. In general, the overall system performance depends on the allocation of delay budget among each delay module given in end-to-end delay as shown in equation (2).

A transverse view of the 3-D Pareto surface, i.e., R-D curve under multiple d values is shown in Figure 3, where d is measured in ratio of ME (Motion Estimation) time to a frame interval. For example, d =0% indicates that no ME is conducted and d = 100% indicates that ME takes the entire frame interval. Recall that previous works on rate control show only one R-D curves of the similar shape as in Figure 3. This is because they always assume a fixed but unspecified encoding time d in their R-D models, which is a special case of d-R-D model. If we provide higher rate to source bits, the channel errors will destroy the signal. On the other hand, if use we reduce the rate given to the source bits, the reconstruction quality will be poor because of over quantization. By these conflicting aspects, there exists a trade-off. Hence rate allocation plays a very important role in developing high-performance end-to-end systems.

Figure 3. Source Distortion D versus Source Rate R, at Multiple Source Encoding time d

It can be seen that with adaptive rate allocation and control, the video encoder always chooses appropriate source/channel coding bit rates and encoder settings, which yields significantly improved picture quality at the receiver end.

Advanced technologies are providing lower error rates and higher bit rates, but guaranteed quality of service will certainly lead to higher user demands of services for which wireless video applications transliterate to higher resolution images of excellent visual quality.

This paper has surveyed studies that have been done in the area of QoS improvement for wireless video communication. Challenges in achieving optimized video quality and motivation for this objective have been discussed by using a DRD optimized rate control algorithm which minimizes the end-to-end distortion subject to the transmission rate and delay constraints. On the other hand, a d-P-R-D model minimizes the source coding distortion under the constraints of source coding rate, encoding power, and maximum encoding delay. These optimization methods of video communication can gain good QoS and achieves 40% to 60% computation reduction on aspects of video coding. The results are analyzed with respect to the QoS measures of video communication. Hence, providing QoS guarantees for application services over the next-generation wireless networks and a common QoS framework will be proposed for the future research.