Leveraging Frame Aggregation for Estimating WiFi Available Bandwidth
12 Jun 2017-pp 1-9
TL;DR: It is shown not only how frame aggregation breaks existing lightweight mechanisms for link characterization but also how to carefully construct packet sequences that induce frame aggregation to capture the WiFi available bandwidth.
Abstract: WiFi has emerged as a pivotal technology for mobile devices offering the potential for exceptional connectivity speeds. Unfortunately, the performance of WiFi may vary significantly making WiFi link characterization (and more broadly the path characterization) an essential element of the user Quality of Experience (QoE). The key challenge that emerges with respect to link characterization is how to characterize performance in an efficient manner. In this paper, we explore how the existing frame aggregation mechanisms introduced by 802.11e can be leveraged to achieve such a goal. We show not only how frame aggregation breaks existing lightweight mechanisms for link characterization but also how to carefully construct packet sequences that induce frame aggregation to capture the WiFi available bandwidth. We construct a proof of concept system, AIWC (Aggregation Intensity based Wifi Characterization), to demonstrate the aforementioned concepts with significant improvements versus prior work.
Citations
More filters
01 Dec 2019
TL;DR: This paper considers cost-aware edge resource probing (CERP) framework design for infrastructure-free edge computing wherein a task device self-organizes its resource probing for informed computation offloading and devise a data-driven layered learning mechanism for more practical and complicated application environments.
Abstract: To meet the stringent requirement of artificial intelligence applications, such as face recognition and video streaming analytics, a resource-constrained device can offload its task to nearby resource-rich devices in edge computing. Resource awareness, as a prime prerequisite for offloading decision-making, is critical for achieving efficient collaborative computation performance. In this paper, we consider cost-aware edge resource probing (CERP) framework design for infrastructure-free edge computing wherein a task device self-organizes its resource probing for informed computation offloading. We first propose a multi-stage optimal stopping formulation for the problem, and derive the optimal probing strategy which reveals a nice multi-threshold structure. Accordingly, we then devise a data-driven layered learning mechanism for more practical and complicated application environments. Layered learning enables the task device to adaptively learn the optimal probing sequence and decision thresholds at runtime, aiming at deriving a good balance between the gain of choosing the best edge device and the accumulated cost of deep resource probing. We further conduct thorough performance evaluation of the proposed CERP schemes using both extensive numerical simulations and realistic system prototype implementation, which demonstrate the superior performance of CERP in the diverse application scenarios.
21 citations
Cites background from "Leveraging Frame Aggregation for Es..."
...2, one resource probing operation can incur significant energy cost as well as time overhead, since one-shot probing operation typically involves wireless connection establishment, bandwidth measurement [14] [15], CPU resource profiling/prediction, and resource information feedback....
[...]
TL;DR: This work proposes an adaptive ML-based approach for frame size selection on a per-user basis by taking into account both specific channel conditions and global performance indicators, and relies on standard frame aggregation mechanisms.
Abstract: Software-Defined Networking (SDN) is gaining a lot of traction in wireless systems with several practical implementations and numerous proposals being made. Despite instigating a shift from monolithic network architectures towards more modulated operations, automated network management requires the ability to extract, utilise and improve knowledge over time. Beyond simply scrutinizing data, Machine Learning (ML) is evolving from a simple tool applied in networking to an active component in what is known as Knowledge-Defined Networking (KDN). This work discusses the inclusion of ML techniques in the specific case of Software-Defined Wireless Local Area Networks (SD-WLANs), paying particular attention to the frame length optimization problem. With this in mind, we propose an adaptive ML-based approach for frame size selection on a per-user basis by taking into account both specific channel conditions and global performance indicators. By relying on standard frame aggregation mechanisms, the model can be seamlessly embedded into any Enterprise SD-WLAN by obtaining the data needed from the control plane, and then returning the output back to this in order to efficiently adapt the frame size to the needs of each user. Our approach has been gauged by analysing a multitude of scenarios, with the results showing an average improvement of 18.36% in goodput over standard aggregation mechanisms.
12 citations
Cites background from "Leveraging Frame Aggregation for Es..."
...Among the factors determining such channel conditions, many works have proved the greater role of the PHY rate with respect to others [20, 21, 73]....
[...]
Proceedings Article•
01 Nov 2018
TL;DR: Results show that in suitable conditions ABw estimates can be used as a proxy for TCP achievable throughput measurements, while generating much lower traffic volumes (a critical asset on MBB networks), and that policies enforced by service providers may significantly alter the difference between the two estimated metrics.
Abstract: Mobile broadband (MBB) networks are being increasingly used worldwide, and they are planned to steadily evolve to support new services, bigger user base, and booming machine-to-machine communications. In this scenario, performance measurements on deployed networks becomes crucial. In particular, being aware of the achievable throughput has multiple important uses ranging from path and server selection, to root-cause analysis. Throughput estimation suffers from high intrusiveness and dependence on transport and application protocol, while Available Bandwidth is a network-layer metric characterizing the spare capacity of path, not suffering from any of these shortcomings. However, ABw estimation tools have been mainly developed focusing on wired networks, with limited attention to mobile and wireless scenarios, multi-homed mobile nodes, and their peculiarities.In this work, we analyze ABw estimates and TCP achievable throughput measurements on a real 3G/4G testbed (the MONROE platform) performing tests with different providers from countries across Europe. The two metrics are compared for results and in terms of measurement intrusiveness and time cost, highlighting the non-trivial relationship between them. In particular, results show that in suitable conditions ABw estimates can be used as a proxy for TCP achievable throughput measurements, while generating much lower traffic volumes (a critical asset on MBB networks), and that policies enforced by service providers may significantly alter the difference between the two estimated metrics. Future research on how to exploit these findings is discussed as well. We published as open data packet traces and logs of the measurements.
7 citations
Cites background from "Leveraging Frame Aggregation for Es..."
...Song and Striegel in [15] further analyze this mechanism, and leverage it by inducing frame aggregation on probe packets to estimate ABw with higher accuracy....
[...]
15 Jun 2020
TL;DR: This article proposes an analytical model based on a Markov chain which estimates the theoretical aggregation level for different amounts of cross traffic and shows that the theoretical model gives an accurate estimation of the frame aggregation level and that it can be used to infer the network load.
Abstract: Over the past two decades, Wi-Fi technology (defined by the IEEE 802.11 standard) has become a prominent wireless network access technology. In many situations, a device may attach to several Wi-Fi access points within the radio range. The operating system makes its choice over metrics that do not take into account the actual available capacity. To fill this gap, several proposals have been made to infer capacity, for example by estimating the occupied proportion of the channel. However, these techniques are being thwarted by the mechanisms recently introduced in 802.11 to improve transmission speeds, in particular frame aggregation. In this article, we focus on busy time inference based on frame aggregation level. We propose an analytical model based on a Markov chain which estimates the theoretical aggregation level for different amounts of cross traffic. We validate its accuracy against simulations carried out on the ns-3 network simulator and an ad-hoc simulator. Results show that the theoretical model gives an accurate estimation of the frame aggregation level and that it can be used to infer the network load.
6 citations
Cites background or methods from "Leveraging Frame Aggregation for Es..."
...To detect an aggregation, both WBest+ and AIWC follow a threshold-based method: if an inter-arrival time is below a given threshold, the two packets are considered as aggregated....
[...]
...AIWC [7] estimates the frame aggregation level at the receiver to measure link congestion and deduce the available bandwidth....
[...]
...If this time is below a given threshold (set as 300μs in [6] and 400μs in [7]), the two packets are considered as aggregated....
[...]
...We have tested the threshold-based method introduced in [6], [7] to estimate the aggregation level....
[...]
TL;DR: A novel passive client-side approach that delivers efficient and accurate characterization by taking advantage of the properties of frame aggregation (FA) and block acknowledgment (BA) thus making it feasible to deploy on IoT devices that have limited computation power.
Abstract: The advancement of the Internet of Things (IoT) is bringing unprecedented convenience into our daily life. However, with the relentlessly increasing number of mobile devices connected to the Internet, the wireless network environment is becoming more crowded than ever before. Particularly, WiFi, with its evolving role in IoT, is shouldering a tremendous amount of traffic from IoT and other mobile devices. As a result, exploding numbers of competing devices, encroachment by cellular technology, and dramatic increases in content richness deliver a more variable Quality of Experience (QoE) on WiFi than desired. Moreover, such variance tends to occur both across time and space making it an extremely difficult problem to debug. Existing active approaches tend to be expensive or impractical while existing passive approaches tend to be too narrow. To conduct efficient and nonobtrusive WiFi traffic characterization, in this article, we propose a novel passive client-side approach that delivers efficient and accurate characterization by taking advantage of the properties of frame aggregation (FA) and block acknowledgment (BA). The devised approach requires only capturing and analyzing certain types of control packets thus making it feasible to deploy on IoT devices that have limited computation power. We show in this article that we can accurately derive important characterization metrics, such as airtime, queuing information, and transmission rates with only a minimal amount of observed BAs. We show through extensive experiments the validity of our approach and conduct validation studies in the dense environment of a campus tailgate.
5 citations
References
More filters
01 Apr 2003
TL;DR: This paper presents pathChirp, a new active probing tool for estimating the available bandwidth on a communication network path based on the concept of ''self-induced congestion,'' which features an exponential flight pattern of probes the authors call a chirp.
Abstract: This paper presents pathChirp, a new active probing tool for estimating the available bandwidth on a communication network path. Based on the concept of ''self-induced congestion,'' pathChirp features an exponential flight pattern of probes we call a chirp. Packet chips offer several significant advantages over current probing schemes based on packet pairs or packet trains. By rapidly increasing the probing rate within each chirp, pathChirp obtains a rich set of information from which to dynamically estimate the available bandwidth. Since it uses only packet interarrival times for estimation, pathChirp does not require synchronous nor highly stable clocks at the sender and receiver. We test pathChirp with simulations and Internet experiments and find that it provides good estimates of the available bandwidth while using only a fraction of the number of probe bytes that current state-of-the-art techniques use.
822 citations
"Leveraging Frame Aggregation for Es..." refers background or methods in this paper
...We begin by comparing the available bandwidth accuracy of AIWC versus prior AB estimation works, including PathChirp [4], Spruce [11], and WBest+ [8]....
[...]
...Bandwidth Estimation: The notion of end-to-end bandwidth estimation can be reduced to the problem of trying to predict what performance can be achieved across said path [3], [4], [6], [19], [20]....
[...]
...We show significant performance improvements versus the existing body of AB literature [4], [8], [11]....
[...]
...The notion of PRM [2], [4]–[6], [13] is based on the concept of self-induced congestion....
[...]
...As mentioned earlier, the works in the PRM (Packet Rate Model) approach (PathLoad [6], PathChirp [4], and TOPP [5]) and the PGM (Packet Gap Model) approach (IGI [3], Spruce [11] and Abing [14]) were primarily focused on wired networks....
[...]
TL;DR: This work describes an end-to-end methodology, called self-loading periodic streams (SLoPS), for measuring avail-bw, and uses pathload, a nonintrusive tool, to evaluate the variability ("dynamics") of the avail- bw in Internet paths.
Abstract: The available bandwidth (avail-bw) in a network path is of major importance in congestion control, streaming applications, quality-of-service verification, server selection, and overlay networks. We describe an end-to-end methodology, called self-loading periodic streams (SLoPS), for measuring avail-bw. The basic idea in SLoPS is that the one-way delays of a periodic packet stream show an increasing trend when the stream's rate is higher than the avail-bw. We have implemented SLoPS in a tool called pathload. The accuracy of the tool has been evaluated with both simulations and experiments over real-world Internet paths. Pathload is nonintrusive, meaning that it does not cause significant increases in the network utilization, delays, or losses. We used pathload to evaluate the variability ("dynamics") of the avail-bw in Internet paths. The avail-bw becomes significantly more variable in heavily utilized paths, as well as in paths with limited capacity (probably due to a lower degree of statistical multiplexing). We finally examine the relation between avail-bw and TCP throughput. A persistent TCP connection can be used to measure roughly the avail-bw in a path, but TCP saturates the path and increases significantly the path delays and jitter.
765 citations
27 Oct 2003
TL;DR: Spruce as mentioned in this paper is a light-weight tool for measuring available bandwidth, and compares it with two existing tools, IGI and Pathload, over 400 different Internet paths, focusing on accuracy, failure patterns, probe overhead, and implementation issues.
Abstract: Available bandwidth estimation is useful for route selection in overlay networks, QoS verification, and traffic engineering. Recent years have seen a surge in interest in available bandwidth estimation. A few tools have been proposed and evaluated in simulation and over a limited number of Internet paths, but there is still great uncertainty in the performance of these tools over the Internet at large.This paper introduces Spruce, a simple, light-weight tool for measuring available bandwidth, and compares it with two existing tools, IGI and Pathload, over 400 different Internet paths. The comparison focuses on accuracy, failure patterns, probe overhead, and implementation issues. The paper verifies the measured available bandwidth by comparing it to Multi-Router Traffic Grapher (MRTG) data and by measuring how each tool responds to induced changes in available bandwidth.The measurements show that Spruce is more accurate than Pathload and IGI. Pathload tends to overestimate the available bandwidth whereas IGI becomes insensitive when the bottleneck utilization is large.
690 citations
TL;DR: This paper uses modeling, measurements, and simulations to better characterize the interaction between probing packets and the competing network traffic, and presents two available bandwidth measurement techniques, the initial gap increasing (IGI) method and the packet transmission rate (PTR) method.
Abstract: The packet pair mechanism has been shown to be a reliable method to measure the bottleneck link capacity on a network path, but its use for measuring available bandwidth is more challenging. In this paper, we use modeling, measurements, and simulations to better characterize the interaction between probing packets and the competing network traffic. We first construct a simple model to understand how competing traffic changes the probing packet gap for a single-hop network. The gap model shows that the initial probing gap is a critical parameter when using packet pairs to estimate available bandwidth. Based on this insight, we present two available bandwidth measurement techniques, the initial gap increasing (IGI) method and the packet transmission rate (PTR) method. We use extensive Internet measurements to show that these techniques estimate available bandwidth faster than existing techniques such as Pathload, with comparable accuracy. Finally, using both Internet measurements and ns simulations, we explore how the measurement accuracy of active probing is affected by factors such as the probing packet size, the length of probing packet train, and the competing traffic on links other than the tight link.
689 citations
"Leveraging Frame Aggregation for Es..." refers background or methods in this paper
...Bandwidth Estimation: The notion of end-to-end bandwidth estimation can be reduced to the problem of trying to predict what performance can be achieved across said path [3], [4], [6], [19], [20]....
[...]
..., [3], [11], [14]) exploits the gap information of the received packet (Grcv) to infer the cross traffic....
[...]
...As mentioned earlier, the works in the PRM (Packet Rate Model) approach (PathLoad [6], PathChirp [4], and TOPP [5]) and the PGM (Packet Gap Model) approach (IGI [3], Spruce [11] and Abing [14]) were primarily focused on wired networks....
[...]
19 Aug 2002
TL;DR: An end-to-end methodology, called Self-Loading Periodic Streams (SLoPS), for measuring avail-bw, and the relation between avail- bw and TCP throughput is examined.
Abstract: The available bandwidth (avail-bw) in a network path is of major importance in congestion control, streaming applications, QoS verification, server selection, and overlay networks. We describe an end-to-end methodology, called Self-Loading Periodic Streams (SLoPS), for measuring avail-bw. The basic idea in SLoPS is that the one-way delays of a periodic packet stream show an increasing trend when the stream's rate is higher than the avail-bw. We implemented SLoPS in a tool called pathload. The accuracy of the tool has been evaluated with both simulations and experiments over real-world Internet paths. Pathload is non-intrusive, meaning that it does not cause significant increases in the network utilization, delays, or losses. We used pathload to evaluate the variability ('dynamics') of the avail-bw in some paths that cross USA and Europe. The avail-bw becomes significantly more variable in heavily utilized paths, as well as in paths with limited capacity (probably due to a lower degree of statistical multiplexing). We finally examine the relation between avail-bw and TCP throughput. A persistent TCP connection can be used to roughly measure the avail-bw in a path, but TCP saturates the path, and increases significantly the path delays and jitter.
611 citations
"Leveraging Frame Aggregation for Es..." refers background in this paper
...The notion of PRM [2], [4]–[6], [13] is based on the concept of self-induced congestion....
[...]