kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Machine Learning for Wireless Link Adaptation: Supervised and Reinforcement Learning Theory and Algorithms
KTH, School of Electrical Engineering and Computer Science (EECS), Intelligent systems, Information Science and Engineering.ORCID iD: 0000-0001-7974-5096
2021 (English)Doctoral thesis, comprehensive summary (Other academic)
Abstract [en]

Wireless data communication is a complex phenomenon. Wireless links encounter random, time-varying, channel effects that are challenging to predict and compensate. Hence, to optimally utilize the channel, wireless links adapt the data transmission parameters in real time. This process, known as wireless link adaptation, can lead to large gains in link performance. Link adaptation is hence an integral part of state-of-the-art wireless deployments.

Existing link adaptation schemes use simple heuristics that match the data transmission rate to the estimated channel. These schemes have proven to be useful for the ubiquitous wireless services of voice telephony and mobile broadband. However, as wireless networks increase in complexity and also evolve to support new service types, these link adaptation schemes are rapidly becoming inadequate. The reason for this change is threefold: first, in several operating scenarios, simple heuristics-based link adaptation does not fully exploit the available channel. Second, the heuristics are typically tuned empirically for good performance, which incurs additional expense and can be error-prone. Finally, traditional link adaptation does not naturally extend to applications beyond the traditional wireless services, for example to industrial control or vehicular communications.

In this thesis, we address wireless link adaptation through machine learning. Our proposed solutions efficiently navigate the link parameter space by learning from the available information. These solutions thus improve the link performance compared to the state-of-the-art, for example by doubling the link throughput. Further, we advance link adaptation support for new wireless services by optimizing the link for complex performance objectives. Finally, we also introduce mechanisms that autonomously tune the link adaptation parameters with respect to the operating environment. Our schemes hence mitigate the dependence on empirical configurations adopted in current wireless networks.

This thesis is composed of six technical papers. Based on these papers, there are three key contributions of this thesis: a neural link adaptation model (Paper I, Paper II, and Paper III), link adaptation under packet error rate constraints (Paper IV  and Paper V), and efficient model-based link adaptation (Paper VI).

In this thesis, we emphasise the theoretical underpinnings of our proposed machine learning schemes for link adaptation. We approach this goal in three ways: First, we make theoretically reasoned choices for machine learning models and learning algorithms for link adaptation. Second, we extend these models for the specific problem formulations encountered in link adaptation. For this, we develop rigorous problem formulations that are analyzed using classical techniques. Third, we develop theoretical results for the real-time behaviour of the proposed schemes. These bounds extend the machine learning state-of-the-art in terms of performance bounds for stochastic online optimization. The contributions of this thesis hence go beyond the realm of wireless optimization, and extend to new developments applicable to broader machine learning problems. 

Abstract [sv]

Trådlös datakommunikation är ett komplext fenomen. Trådlösa länkar stöter på slumpmässiga och tidsvarierande kanaleffekter som är utmanande att förutsäga och kompensera för. För att optimalt utnyttja den trådlösa kanalen anpassar därför kommunikationssystem dataöverföringsparametrarna i realtid. Denna process, även kallad trådlös länkanpassning, kan leda till stora vinster i länkprestanda. Länk-anpassning är därför en integrerad del av alla moderna kommunikationssystem. Befintliga metoder för länkanpassning använder enkla heuristiker som anpassar dataöverföringshastigheten till den skattade trådlösa kanalen. Dessa system har visat sig vara användbara för de brett använda trådlösa tjänsterna rösttelefoni och mobilt bredband. Eftersom trådlösa nätverk ökar i komplexitet och också utvecklas för att stödja nya tjänstetyper, blir dock dessa metoder för länkanpassning snabbt otillräckliga. Anledningen till detta är trefaldig: För det första så utnyttjar heuristikbaserad länkanpassning i flera nya tjänster utnyttjar helt enkelt inte den tillgängliga kanalen till fullo. För det andra så är heuristiken vanligtvis anpassad empiriskt för bra prestanda, vilket kan vara felbenäget i nya scenarion och vilket medför extra kostnader. Slutligen så generaliserar traditionell länkanpassning inte naturligt till tillämpningar som går utöver de traditionella trådlösa tjänsterna, till exempel till industriella reglersystem eller fordonskommunikation. I denna avhandling behandlar vi länkanpassning genom maskininlärning. Våra föreslagna system utforskar effektivt länkparameterutrymmet genom att lära av tillgänglig information. De föreslagna metoderna förbättrar således länkprestandan jämfört med den senaste tekniken, till exempel genom att fördubbla länkgenomströmningen. Vidare utvecklar vi också länkadaptationsstöd för nya trådlösa tjänster genom att optimera länken för mer komplexa prestandamål. Slutligen så introducerar vi också mekanismer som autonomt justerar länkanpassningsparametrarna baserat på driftsmiljön. Våra system mildrar därmed beroendet på empiriska konfigurationer som används i nuvarande trådlösa nätverk. Denna avhandling består av sex tekniska artiklar. Baserat på dessa artiklar finns det tre viktiga bidrag från denna avhandling: En modell för anpassning av neurala länkar (Paper I, Paper II och Paper III), länkanpassning under begränsningar i paketfelfrekvensen (Paper IV och Paper V), och effektiv modellbaserad länkanpassning (Paper VI). I denna avhandling betonar vi den teoretiska grunden för våra föreslagna maskininlärningsmetoder för länkanpassning. Vi närmar oss detta mål på tre sätt: För det första gör vi teoretiskt motiverade val för maskininlärningsmodeller och inlärningsalgoritmer för länkanpassning. För det andra utökar vi dessa modeller för de specifika problemformuleringar som påträffas vid länkanpassning. För detta utvecklar vi noggranna problemformuleringar som analyseras med klassiska tekniker. För det tredje utvecklar vi teoretiska resultat för de föreslagna systemens realtidsbeteende. Dessa gränser utökar fältet maskininlärningen när det gäller prestationsgränser för stokastisk online-optimering. Bidragen från denna avhandling går alltså utöver området för trådlös kommunikation och sträcker sig till nya tillämpningsområden.

Abstract [hi]

वायरलेस डेटा संचार एक जटिल प्रक्रिया है। वायरलेस कड़ियाँ  (लिंक्स) अव्यवस्थित और क्रम-रहित चैनल प्रभावों का सामना करतीं हैं, जिनकी क्षतिपूर्ति कर पाना चुनौतीपूर्ण है। अतः, चैनल का सर्वोत्तम  उपयोग करने के लिए, वायरलेस लिंक्स वास्तविक समय में डेटा संचारण मापदंडों (पैरामीटर्स) को अनुकूलित करते हैं। इस प्रक्रिया को वायरलेस लिंक अनुकूलन के नाम से जाना जाता है, जो अत्याधुनिक वायरलेस परिनियोजन का एक अभिन्न अंग है।

मौजूदा लिंक अनुकूलन योजनाएं अनुभव पर आधारित, सरल, अनुमानों का उपयोग करती हैं। आमतौर से, ये योजनांए डेटा संचारण दर का अनुमानित वायरलेस चैनल से मेल कराती हैं  । पूर्वकाल में, ये योजनाएं दूरभाष और मोबाइल ब्रॉडबैंड की सर्वव्यापी वायरलेस सेवाओं के लिए उपयोगी साबित हुई हैं। किन्तु, जैसे-जैसे वायरलेस नेटवर्क जटिल होते जा रहे हैं , और नए प्रकार की संचारण -व्यवस्थाएं विकसित हो रही हैं, मौजूदा लिंक अनुकूलन योजनाएं भी तेजी से अपर्याप्त होती चली जा रही हैं। इस परिवर्तन के यह तीन मुख्य कारण हैं: पहला, कई परिदृश्यों में, सरल लिंक अनुकूलन मौजूदा चैनल का पूरी तरह से उपयोग नहीं कर पाता। दूसरा, संचारण मापदंडों को सामान्यतः आनुभाविक रूप से चुना जाता है, जो अतिरिक्त संचरण को बढ़ाता है और इसमें त्रुटियों की सम्भावना अधिक होती  है। अंततः,पारंपरिक लिंक अनुकूलन नयी सेवा-प्रयोगों की ओर स्वाभाविक रूप से विस्तार नहीं करता - उदाहरण के लिए, औद्योगिक नियंत्रण अथवा  वाहन-आधारित संचार।

इस शोध प्रबंध (थीसिस) में, हम मशीन लर्निंग के माध्यम से वायरलेस लिंक अनुकूलन का अनुसंधान करते हैं। हमारे प्रस्तावित समाधान सामान्य संचारण जानकारी से सीखकर, लिंक संचरण मापदंडों का स्वतः और  कुशलतापूर्वक सञ्चालन करते हैं। अत्याधुनिक अनुकूलन विधियों  की तुलना में, हमारे  समाधान लिंक निष्पादन (परफॉरमेंस) में सुधार करते हैं, उदाहरण के लिए लिंक प्रवाह क्षमता (थ्रूपुट) को दोगुना करके। इसके अतिरिक्त,हमारे समाधान  जटिल निष्पादन उद्देश्यों के लिए लिंक को अनुकूलित करके नई वायरलेस सेवाओं को लाभ पहुंचाते हैं। अंत में, हम ऐसी तकनीक भी प्रस्तुत  करते हैं जो वायरलेस वातावरण के आधार पर, लिंक अनुकूलन मापदंडों को स्वतः संचालित करती है। इस प्रकार, हमारी प्रस्तावित योजनाएं आज के वायरलेस नेटवर्क की  अनुभवजन्य निर्भरता कम करती हैं। 

इस थीसिस में छह तकनीकी पत्र समाहित हैं। इन पत्रों के आधार पर,यह थीसिस तीन प्रमुख क्षेत्रों में योगदान देती है: एक न्यूरल लिंक अनुकूलन मॉडल (पेपर I, पेपर II और पेपर III), पैकेट त्रुटि दर की कमी के तहत लिंक अनुकूलन (पेपर IV और पेपर V), और मॉडल आधरित कुशल लिंक अनुकूलन (पेपर VI)।

इस थीसिस में, हम लिंक अनुकूलन के लिए अपनी प्रस्तावित मशीन लर्निंग योजनाओं की सैद्धांतिक मजबूती  पर बल देते हैं। इस लक्ष्य तक पहुँचने के लिए हम निम्न तीन सूत्रों को अपनाते हैं: सबसे पहले, हम लिंक अनुकूलन के दृष्टिकोण से उचित, मशीन लर्निंग के सैद्धांतिक मॉडल्स और अल्गोरिथम्स का प्रयोग करते हैं। दूसरा, हम लिंक अनुकूलन में आई विशेष समस्याओं के हेतु मशीन लर्निंग तकनीकों का विस्तार करते हैं। तीसरा, हम प्रस्तावित योजनाओं के वास्तविक-समय व्यवहार  के लिए सैद्धांतिक परिणाम विकसित करते हैं। ये परिणाम निष्पादन सीमा के संदर्भ में अत्याधुनिक मशीन लर्निंग को भी विकसित करती हैं। अतः इस शोध प्रबंध के योगदान वायरलेस अनुकूलन से बढ़कर, मशीन लर्निंग में व्याप्त समस्याओं पर नए विकास की ओर बढ़ाव

Place, publisher, year, edition, pages
Stockholm: KTH Royal Institute of Technology, 2021. , p. 166
Series
TRITA-EECS-AVL ; 2021:35
Keywords [en]
Wireless Communications, Reinforcement Learning, Multi-Armed Bandits, Thompson Sampling, Convex Optimization, Deep Learning
National Category
Communication Systems Telecommunications
Research subject
Electrical Engineering
Identifiers
URN: urn:nbn:se:kth:diva-293545ISBN: 978-91-7873-886-1 (print)OAI: oai:DiVA.org:kth-293545DiVA, id: diva2:1548049
Public defence
2021-05-20, F3, Lindstedtsvägen 26, Stockholm, 13:00 (English)
Opponent
Supervisors
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)
Note

QC 20210505

Available from: 2021-05-05 Created: 2021-04-28 Last updated: 2022-06-25Bibliographically approved
List of papers
1. Deep learning for frame error probability prediction in bicm-ofdm systems
Open this publication in new window or tab >>Deep learning for frame error probability prediction in bicm-ofdm systems
2018 (English)In: 2018 ieee international conference on acoustics, speech and signal processing (icassp), IEEE, 2018, p. 6658-6662Conference paper, Published paper (Refereed)
Abstract [en]

In the context of wireless communications, we propose a deep learning approach to learn the mapping from the instantaneous state of a frequency selective fading channel to the corresponding frame error probability (FEP) for an arbitrary set of transmission parameters. We propose an abstract model of a bit interleaved coded modulation (BICM) orthogonal frequency division multiplexing (OFDM) link chain and show that the maximum likelihood (ML) estimator of the model parameters estimates the true FEP distribution. Further, we exploit deep neural networks as a general purpose tool to implement our model and propose a training scheme for which, even while training with the binary frame error events (i.e., ACKs/NACKs), the network outputs converge to the FEP conditioned on the input channel state. We provide simulation results that demonstrate gains in the FEP prediction accuracy with our approach as compared to the traditional effective exponential SIR metric (EESM) approach for a range of channel code rates, and show that these gains can be exploited to increase the link throughput.

Place, publisher, year, edition, pages
IEEE, 2018
Keywords
FEP, BICM-OFDM, Deep Learning, Neural Networks, Link Adaptation
National Category
Telecommunications
Identifiers
urn:nbn:se:kth:diva-237157 (URN)10.1109/ICASSP.2018.8461864 (DOI)000446384606163 ()2-s2.0-85054259851 (Scopus ID)
Conference
2018 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP)
Funder
Wallenberg Foundations
Note

QC 20181025

Available from: 2018-10-25 Created: 2018-10-25 Last updated: 2022-06-26Bibliographically approved
2. A Learning Approach for Optimal Codebook Selection in Spatial Modulation Systems
Open this publication in new window or tab >>A Learning Approach for Optimal Codebook Selection in Spatial Modulation Systems
Show others...
2018 (English)In: 2018 conference record of 52nd asilomar conference on signals, systems, and computers / [ed] Matthews, M B, IEEE , 2018, p. 1800-1804Conference paper, Published paper (Refereed)
Abstract [en]

For spatial modulation (SM) systems that utilize multiple transmit antennas/patterns with a single radio front-end, we propose a learning approach to predict the average symbol error rate (SER) conditioned on the instantaneous channel state. We show that the predicted SER can he used to lower the average SER over Rayleigh fading channels by selecting the optimal codebook in each transmission instance. Further by exploiting that feedforward artificial neural networks (ANNs) trained with a mean squared error (MSE) criterion estimate the conditional a posteriori probabilities, we maximize the expected rate for each transmission instance and thereby improve the link spectral efficiency.

Place, publisher, year, edition, pages
IEEE, 2018
Series
Conference Record of the Asilomar Conference on Signals Systems and Computers, ISSN 1058-6393
National Category
Computer and Information Sciences
Identifiers
urn:nbn:se:kth:diva-252674 (URN)10.1109/ACSSC.2018.8645407 (DOI)000467845100317 ()2-s2.0-85062983648 (Scopus ID)978-1-5386-9218-9 (ISBN)
Conference
52nd Asilomar Conference on Signals, Systems, and Computers, OCT 28-NOV 01, 2018, Pacific Grove, CA
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)
Note

QC 20190603

Available from: 2019-06-03 Created: 2019-06-03 Last updated: 2022-06-26Bibliographically approved
3. Contextual multi-armed bandits for link adaptation in cellular networks
Open this publication in new window or tab >>Contextual multi-armed bandits for link adaptation in cellular networks
Show others...
2019 (English)In: NetAI 2019 - Proceedings of the 2019 ACM SIGCOMM Workshop on Network Meets AI and ML, Part of SIGCOMM 2019, Association for Computing Machinery (ACM), 2019, p. 44-49Conference paper, Published paper (Refereed)
Abstract [en]

Cellular networks dynamically adjust the transmission parameters for a wireless link in response to its time-varying channel state. This is known as link adaptation, where the typical goal is to maximize the link throughput. State-of-the-art outer loop link adaptation (OLLA) selects the optimal transmission parameters based on an approximate, offline, model of the wireless link. Further, OLLA refines the offline model by dynamically compensating any deviations from the observed link performance. However, in practice, OLLA suffers from slow convergence and a sub-optimal link throughput. In this paper, we propose a link adaptation approach that overcomes the shortcomings of OLLA through a novel learning scheme. Our approach relies on contextual multi-armed bandits (MAB), where the context vector is composed of the instantaneous wireless channel state along with side information about the link. For a given context, our approach learns the success probability for each of the available transmission parameters, which is then exploited to select the throughput-maximizing parameters. Through numerical experiments, we show that our approach converges faster than OLLA and achieves a higher steady-state link throughput. For frequent and infrequent channel reports respectively, our scheme outperforms OLLA by 15% and 25% in terms of the steady-state link throughpu.

Place, publisher, year, edition, pages
Association for Computing Machinery (ACM), 2019
Keywords
Artificial neural networks, Cellular networks, Contextual multiarmed bandits, Outer loop link adaptation
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
urn:nbn:se:kth:diva-262526 (URN)10.1145/3341216.3342212 (DOI)000506862100007 ()2-s2.0-85072036655 (Scopus ID)
Conference
2019 ACM SIGCOMM Workshop on Network Meets AI and ML, NetAI 2019, Part of SIGCOMM 2019, 23 August 2019
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)
Note

QC 20191028

Part of ISBN 9781450368728

Available from: 2019-10-28 Created: 2019-10-28 Last updated: 2024-10-28Bibliographically approved
4. Bayesian Link Adaptation under a BLER Target
Open this publication in new window or tab >>Bayesian Link Adaptation under a BLER Target
2020 (English)In: 21st IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), 2020, Institute of Electrical and Electronics Engineers (IEEE), 2020, p. 2735-2740, article id 9147432Conference paper, Published paper (Refereed)
Abstract [en]

The optimal modulation and coding scheme (MCS) for wireless transmission depends on the dynamic wireless channel state. Hence, wireless link adaptation relies on periodically reported channel quality index (CQI) values to select the optimal MCS for each transmission instance. However, to optimize link performance for a given wireless environment, current link adaptation techniques rely on tuning parameters such as a block error rate (BLER) target and algorithm adjustments that are difficult to optimize heuristically. Here, we propose BayesLA, a Bayesian link adaptation scheme that does not require careful parameter tuning for optimal link performance in diverse wireless environments. BayesLA, which is inspired by the Thompson Sampling approach widely used for online learning, efficiently learns the MCS success probabilities conditioned on the reported CQI values. Through numerical simulations for a Rayleigh fading wireless channel and a typical cellular link configuration, we demonstrate that BayesLA outperforms state-of-the-art outer loop link adaptation (OLLA) approach in terms of the realized link throughput for a given BLER target.

Place, publisher, year, edition, pages
Institute of Electrical and Electronics Engineers (IEEE), 2020
Series
IEEE International Workshop on Signal Processing Advances in Wireless Communications, ISSN 2325-3789
Keywords
Link Adaptation, Bayesian Statistics, Online Learning, Thompson Sampling, BLER Target
National Category
Communication Systems
Research subject
Information and Communication Technology
Identifiers
urn:nbn:se:kth:diva-273708 (URN)10.1109/SPAWC48557.2020.9154253 (DOI)000620337500051 ()2-s2.0-85090386104 (Scopus ID)
Conference
21st IEEE International Workshop on Signal Processing Advances in Wireless Communications (SPAWC), May 26-29, 2020, Virtual Conference
Projects
European Research Council project AGNOSTIC (742648)
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)
Note

QC 20210401

Available from: 2020-05-25 Created: 2020-05-25 Last updated: 2022-06-26Bibliographically approved
5. Thompson Sampling for Linearly Constrained Bandits
Open this publication in new window or tab >>Thompson Sampling for Linearly Constrained Bandits
2020 (English)In: Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, ML Research Press , 2020, p. 1999-2009Conference paper, Published paper (Refereed)
Abstract [en]

We address multi-armed bandits (MAB) where the objective is to maximize the cumulative reward under a probabilistic linear constraint. For a few real-world instances of this problem, constrained extensions of the well-known Thompson Sampling (TS) heuristic have recently been proposed. However, finite-time analysis of constrained TS is challenging; as a result, only O(sqrt{T}) bounds on the cumulative reward loss (i.e., the regret) are available. In this paper, we describe LinConTS, a TS-based algorithm for bandits that place a linear constraint on the probability of earning a reward in every round. We show that for LinConTS, the regret as well as the cumulative constraint violations are upper bounded by O(log{T}) for the suboptimal arms. We develop a proof technique that relies on careful analysis of the dual problem and combine it with recent theoretical work on unconstrained TS. Through numerical experiments on two real-world datasets, we demonstrate that LinConTS outperforms an asymptotically optimal upper confidence bound (UCB) scheme in terms of simultaneously minimizing the regret and the violation.

Place, publisher, year, edition, pages
ML Research Press, 2020
Series
Proceedings of Machine Learning Research, ISSN 2640-3498 ; 108
National Category
Computer Systems
Research subject
Applied and Computational Mathematics, Mathematical Statistics
Identifiers
urn:nbn:se:kth:diva-273775 (URN)000559931303005 ()2-s2.0-85114758547 (Scopus ID)
Conference
23rd International Conference on Artificial Intelligence and Statistics (AISTATS), 26-28 August 2020
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)
Note

QC 20200602

Available from: 2020-05-29 Created: 2020-05-29 Last updated: 2024-07-11Bibliographically approved
6. Reinforcement Learning for Efficient and Tuning-Free Link Adaptation
Open this publication in new window or tab >>Reinforcement Learning for Efficient and Tuning-Free Link Adaptation
(English)Manuscript (preprint) (Other academic)
Abstract [en]

Wireless links adapt the data transmission parameters to the dynamic channel state -- this is called link adaptation. Classical link adaptation relies on tuning parameters that are challenging to configure for optimal link performance. Recently, reinforcement learning has been proposed to automate link adaptation, where the transmission parameters are modeled as discrete arms of a multi-armed bandit. In this context, we propose a latent learning model for link adaptation that exploits the correlation between data transmission parameters. Further, motivated by the recent success of Thompson sampling for multi-armed bandit problems, we propose a latent Thompson sampling (LTS) algorithm that quickly learns the optimal parameters for a given channel state. We extend LTS to fading wireless channels through a tuning-free mechanism that automatically tracks the channel dynamics. In numerical evaluations with fading wireless channels, LTS improves the link throughout by up to 100% compared to the state-of-the-art link adaptation algorithms.

National Category
Communication Systems
Identifiers
urn:nbn:se:kth:diva-293544 (URN)
Funder
Wallenberg AI, Autonomous Systems and Software Program (WASP)EU, European Research Council, 742648
Note

QC 20210503

Available from: 2021-04-28 Created: 2021-04-28 Last updated: 2022-06-25Bibliographically approved

Open Access in DiVA

fulltext(5532 kB)2299 downloads
File information
File name FULLTEXT01.pdfFile size 5532 kBChecksum SHA-512
13bf491724bbcca7f7083167d201999e73733a711b38e5d387f14ec9d14d0c0f8ac900ce92ffada73c7b42e2e0119b8076f3030b7972f088f8fae1125fb1ac6a
Type fulltextMimetype application/pdf

Other links

zoom link for online defense

Authority records

Saxena, Vidit

Search in DiVA

By author/editor
Saxena, Vidit
By organisation
Information Science and Engineering
Communication SystemsTelecommunications

Search outside of DiVA

GoogleGoogle Scholar
Total: 2300 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

isbn
urn-nbn

Altmetric score

isbn
urn-nbn
Total: 4146 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf