Fig 1 - uploaded by Vaneet Aggarwal
Content may be subject to copyright.
Source publication
The end to end system data performance over a 3G cellular network depends on many factors such as the number of users, interference, multipath propagation, radio resource management techniques as well as the interaction between these mechanisms and the transport protocol's flow and congestion mechanisms. Using controlled experiments in a public cel...
Context in source publication
Similar publications
In beyond 3G networks the user will not be aware of the access network technology used to provide a telecommunications service. Heterogeneous network technologies will be seamlessly integrated in one "common" access network, enabling users to move around and continuously receive their subscribed services. In a commercial environment, this network e...
Citations
... Trends in video content have changed drastically since the advent of social media. The durations of video content from online video-sharing platforms such as YouTube have been drastically reduced over the years [2]. The average duration of the top 10 videos on Facebook was 128 s in 2018 [17]. ...
... The answer to this question is not available in literature. Based on the average duration of movie trailers, the duration of most videos on Facebook, and the average duration of English Premier League (EPL) highlights, we set the maximum duration of a short video to 120 s [1,2,17]. If the video is longer than 120 s, the algorithm runs Subroutine 2 (Long Video Adaptation Algorithm). ...
The traditional client-based HTTP adaptation strategies do not explicitly coordinate between the clients, servers, and cellular networks. A lack of coordination leads to suboptimal user experience. In addition to optimizing Quality of Experience (QoE), other challenges in adapting HTTP adaptive streaming (HAS) to the cellular environment are overcoming unfair allocation of the video rate and inefficient utilization of the bandwidth under the high-dynamics cellular links. Furthermore, the majority of the adaptive strategies ignore important video content characteristics and HAS client information, such as segment duration, buffer size, and video duration, in the video quality selection process. In this paper, we present a content-aware hybrid multi-access edge computing (MEC)-assisted quality adaptation algorithm by taking advantage of the capabilities of edge cloud computing. The proposed algorithm exploits video content characteristics, HAS client settings, and application-layer information to jointly adapt the bitrates of multiple clients. We design separate strategies to optimize the performance of short and long duration videos. We then demonstrate the efficiency of our algorithm against client-based solutions as well as MEC-assisted algorithms. The proposed algorithm guarantees high QoE, equitably selects video rates for clients, and efficiently utilizes the bandwidth for both short and long duration videos. The results from our extensive experiments reveal that the proposed long video adaptation algorithm outperforms state-of-the-art algorithms, with improvements in average video rate, QoE, fairness, and bandwidth utilization of 0.4%–12.3%, 8%–65%, 3.3%–5.7%, and 60%–130%, respectively. Furthermore, when high bandwidth is available to competing clients, the proposed short video adaptation algorithm improves QoE by 11.1% compared to the long video adaptation algorithm.
... It has been shown that they cannot accurately estimate the bandwidth when multiple clients compete for network bottleneck [3,5]. The authors in [13] proposed that multiple clients cannot achieve fair performance when the bottleneck is the air interface. Furthermore, the unfairness increases as the number of competing clients increase. ...
Seamless streaming of high quality video under unstable network condition is a big challenge. HTTP adaptive streaming (HAS) provides a solution that adapts the video quality according to the network conditions. Traditionally, HAS algorithm runs at the client side while the clients are unaware of bottlenecks in the radio channel and competing clients. The traditional adaptation strategies do not explicitly coordinate between the clients, servers, and cellular networks. The lack of coordination has been shown to lead to suboptimal user experience. As a response, multi-access edge computing (MEC)-assisted adaptation techniques emerged to take advantage of computing and content storage capabilities in mobile networks. In this study, we investigate the performance of both MEC-assisted and client-side adaptation methods in a multi-client cellular environment. Evaluation and comparison are performed in terms of not only the video rate and dynamics of the playback buffer but also the fairness and bandwidth utilization. We conduct extensive experiments to evaluate the algorithms under varying client, server, dataset, and network settings. Results demonstrate that the MEC-assisted algorithms improve fairness and bandwidth utilization compared to the client-based algorithms for most settings. They also reveal that the buffer-based algorithms achieve significant quality of experience; however, these algorithms perform poorly compared with throughput-based algorithms in protecting the playback buffer under rapidly varying bandwidth fluctuations. In addition, we observe that the preparation of the representation sets affects the performance of the algorithms, as does the playback buffer size and segment duration. Finally, we provide suggestions based on the behaviors of the algorithms in a multi-client environment.
... Several methods have been used to develop high precision techniques in forecasting 3G network traffic [9] [10] [11]. Reference [10] applied data mining technique in predicting the air interface load of 3G network traffic while reference [12] established that 3G cellular network resource management is influenced by factors such as number of users, multipath propagation, congestion control, transport protocol flow, etc. ...
... In modelling the five-input five-output CANFIS structure, a bell fuzzy axon with the bell-shaped curve as its MF was applied to the input telecommunication network traffic variable, hourly, daily, weekly, monthly and quarterly respectively as shown in equation (12). The fuzzy axon has the advantage of modifying the MF while the network training process continues over back propagation which ensures convergence. ...
... Bell function is given as [20]: (12) where x = input to the node and a1, b1 and c1 are adaptable variables known as premise parameters. ...
Telecommunication network traffic prediction is an important approach that ensure efficient network planning and management. Telecommunication network traffic is univariate and prediction models have mostly been concentrated on single-input and single-output traffic. This study proposes a new approach, the multiple-input multiple-output Coactive Neuro-Fuzzy Inference System (CANFIS) model to predict a five time span univariate hourly, daily, weekly, monthly and quarterly time series of 3G downlink traffic simultaneously. In the modelling process several parameters were used in the configuration of the network. The best model for predicting five-input telecommunication traffic was CANFIS (5-2-5) which employed a Bell membership function, Axon transfer function and Momentum learning rule and the membership function per input of 2. The performance of the model was evaluated by comparing the predicted traffic with actual traffic obtained from a 3G network operator and the results indicate a minimum accuracy measure value of MSE = 0.000486, NRMSE = 0.01120 and percent error = 12.33%.
... In a practical cellular networks, fairness among the equally paying users is an utmost important deployment and optimization criterion. PF is a widely adopted radio resource allocation scheme in access networks [17]. There is no such literature that provides PF-based hybrid precoder for a desired number of RF chains. ...
This paper presents a study on joint radio resource allocation and hybrid precoding in multi-carrier massive multiple-input multiple-output (MIMO) communications for 5G cellular networks. In this paper, we present resource allocation algorithm to maximize the proportional fairness (PF) spectral efficiency under the per subchannel power and the beamforming rank constraints. Two heuristic algorithms are designed. The proportional fairness hybrid beamforming (PFHB) algorithm provides the transmit precoder with proportional fair spectral efficiency among users for the desired number of radiofrequency (RF) chains. Then, we transform the number of RF chains or rank constrained optimization problem into convex semidefinite programming (SDP) problem which can be solved by standard techniques. Inspired by the formulated convex SDP problem; a low complexity, two step, PF relaxed optimization algorithm (PFRO) has been provided for the formulated convex optimization problem. Simulation results show that the proposed suboptimal solution to the relaxed optimization problem is nearoptimal for signal-to-noise ratio SNR 10dB and has a performance gap not greater than 2.33bps/Hz within the SNR range 0 − 25dB. It also outperforms the maximum throughput and PF-based hybrid beamforming schemes for sum spectral efficiency, individual spectral efficiency and fairness index.
... Thus the instable and unfair playback will be inevitably introduced. More unfortunately, measurement studies [9] have shown that the prevalent assumption of TCP fairness may not be true in cellular networks. This indicates that multiple clients would not achieve fair performance even though they somehow learn the channel information and request the true bandwidth share. ...
... It has been widely assumed that TCP can provide fair performance for multiple flows. However, measurement studies over a 3G network [9] has shown that the fairness among multiple competing TCP flows fluctuates significantly when the lasthop cellular link is the bottleneck, i.e., TCP can achieve rather unfair throughput (more than 5 times difference) in cellular networks. The major cause behind this phenomenon is the poor interaction between TCP congestion control and proportional fair (PF) scheduling in cellular networks. ...
In this paper, we present Prius, a hybrid edge cloud and client adaptation framework for HTTP adaptive streaming (HAS) by taking advantage of the new capabilities empowered by recent advances in edge cloud computing. In particular, emerging edge clouds are capable of accessing application-layer and radio access networks (RAN) information in real time. Coupled with powerful computation support, an edge cloud assisted strategy is expected to significantly enrich mobile services. Meanwhile, although HAS has established itself as the dominant technology for video streaming, one key challenge for adapting HAS to mobile cellular networks is in overcoming the inaccurate bandwidth estimation and unfair bitrate adaptation under the highly dynamic cellular links. Edge cloud assisted HAS presents a new opportunity to resolve these issues and achieve systematic enhancement of quality of experience (QoE) and QoE fairness in cellular networks. To explore this new opportunity, Prius overlays a layer of adaptation intelligence at the edge cloud to finalize the adaptation decisions while considering the initial bandwidth-irrelevant bitrate selection at the clients. Prius is able to exploit RAN channel status, client device characteristics as well as application-layer information in order to jointly adapt the bitrate of multiple clients. Prius also adopts a QoE continuum model to track the cumulative viewing experience and an exponential smoothing estimation to accurately estimate future channel under different moving patterns. Extensive trace-driven simulation results show that Prius with hybrid edge cloud and client adaptation is promising under both slow and fast moving environment. Furthermore, Prius adaptation algorithm achieves a near-optimal performance that outperforms exiting strategies.
... Our key performance metrics include video quality metrics such as the streamed video bitrate as selected by the client, the fraction of time for different qualities, the number of quality switches performed by the clients for adaptation, number of video stalls, and stall durations. We also consider network related metrics such as resource utilisation and fairness, which is estimated as [17] ...
... Our work extends their work by investigating the buffer sizes and queuing policies of mobile ISPs, and we found some surprising differences among the three local ISPs. Aggarwal et al. discussed the fairness of 3G networks and found that the fairness of TCP is adversely affected by a mismatch between the congestion control algorithm and the network's scheduling mechanism [3]. A recent study also showed various interesting effects of network protocols and application behaviors on the performance of LTE networks [5]. ...
With the significant increase in cellular data usage, it is critical to better understand the characteristics and behavior of cellular data networks. With both laboratory experiments and crowd-sourcing measurements, we investigated the characteristics of the cellular data networks for the three mobile ISPs in Singapore. We found that i) the transmitted packets tend to arrive in bursts; ii) there can be large variations in the instantaneous throughput over a short period of time; iii) large separate downlink buffers are typically deployed, which can cause high latency when the throughput is low; and iv) the networks typically implement some form of fair queuing policy.
... Figure 2b shows, for each hour of the day, the fraction of near active and idle devices which have changed cell sector since they were last active. We observe that a larger faction of idle devices are using different cell sectors, implying that differences between cell sectors-frequency, back-haul capacity, load, etc. [8]-are a possible cause of the performance differences between active and idle devices. Summary. ...
... An admin may want to run multiple concurrent experiments, so scheduling them such that they do not interfere is important. The bottleneck and complex protocol interactions in cellular networks is typically in the "last mile" [8], so limiting geographic overlap may be sufficient. Incentives for deployment. ...
... Cellular performance studies. Numerous measurement studies of cellular network performance have been conducted, including studies on delay [23,25], TCP performance [24,26], fairness and performance under load [8,36], and application specific performance [19]. While these studies shed light on performance trends, cellular network performance is constantly changing as network providers upgrade and reconfigure their networks. ...
Network service providers, and other parties, require an accurate understanding of the performance cellular networks deliver to users. In particular, they often seek a measure of the network performance users experience solely when they are interacting with their device---a measure we call in-context. Acquiring such measures is challenging due to the many factors, including time and physical context, that influence cellular network performance. This paper makes two contributions. First, we conduct a large scale measurement study, based on data collected from a large cellular provider and from hundreds of controlled experiments, to shed light on the issues underlying in-context measurements. Our novel observations show that measurements must be conducted on devices which (i) recently used the network as a result of user interaction with the device, (ii) remain in the same macro-environment (e.g., indoors and stationary), and in some cases the same micro-environment (e.g., in the user's hand), during the period between normal usage and a subsequent measurement, and (iii) are currently sending/ receiving little or no user-generated traffic. Second, we design and deploy a prototype active measurement service for Android phones based on these key insights. Our analysis of 1650 measurements gathered from 12 volunteer devices shows that the system is able to obtain average throughput measurements that accurately quantify the performance experienced during times of active device and network usage.
The rapid growth of mobile video traffic puts significant pressure on energy drain at the network as well as the end users. Exploiting predicted channel information and designing energy-efficient content delivery protocols has recently drawn attention, which is referred to as predictive, anticipatory, or context-aware resource allocation. In this paper, we investigate how predicted user rates can be exploited for mobile video streaming with the popular HyperText Transfer Protocol (HTTP)-based Adaptive Streaming (HAS) [e.g., dynamic adaptive streaming over HTTP]. To this end, we develop a robust mobile edge cloud assisted stochastic Predictive HTTP Adaptive Streaming (PHAS) optimization framework which utilizes unreliable predictions of wireless data rates in a finite look-ahead window to achieve the following objectives: 1) to obtain an edge-cloud assisted framework for prediction-based HAS and identify its key functional entities and their interactions; 2) to model uncertainty in predicted user rates and propose a robust two-stage QoE optimization approach which dynamically allocates the risks and optimizes system efficiency over a time horizon; 3) to propose a heuristic algorithm allocating time slot ratio for multi-users, which improves network efficiency, fairness and overall QoE under different prediction error variances, wireless link conditions and buffer length constraints. Simulation studies and analytical results show that the proposed solution outperforms traditional methods in terms of average QoE, fairness and energy efficiency.
Mobile network operators have a significant interest in the performance of streaming video on their networks because network dynamics directly influence the Quality of Experience (QoE). However, unlike video service providers, network operators are not privy to the client- or server-side logs typically used to measure key video performance metrics, such as user engagement. To address this limitation, this paper presents the first large-scale study characterizing the impact of cellular network performance on mobile video user engagement from the perspective of a network operator. Our study on a month-long anonymized data set from a major cellular network makes two main contributions. First, we quantify the effect that 31 different network factors have on user behavior in mobile video. Our results provide network operators direct guidance on how to improve user engagement --- for example, improving mean signal-to-interference ratio by 1 dB reduces the likelihood of video abandonment by 2%. Second, we model the complex relationships between these factors and video abandonment, enabling operators to monitor mobile video user engagement in real-time. Our model can predict whether a user completely downloads a video with more than 87% accuracy by observing only the initial 10 seconds of video streaming sessions. Moreover, our model achieves significantly better accuracy than prior models that require client- or server-side logs, yet we only use standard radio network statistics and/or TCP/IP headers available to network operators.
