pith. sign in

arxiv: 1907.01592 · v1 · pith:XUM2POMYnew · submitted 2019-07-02 · 📡 eess.SP

Geolocation of Multiple Noncooperative Emitters Using Received Signal Strength: Sparsity, Resolution, and Detectability

Kurt Bryan , Deborah Walter This is my paper

Pith reviewed 2026-05-25 10:25 UTC · model grok-4.3

classification 📡 eess.SP
keywords geolocationreceived signal strengthsparse approximationcompressed sensingnon-cooperative emittersresolution limitsdetectability
0
0 comments X

The pith

RSS data from multiple indistinguishable emitters yields quantifiable resolution and power detection limits when cast as a sparse recovery problem.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that geolocation of an unknown number of non-cooperative emitters can be performed from received signal strength measurements alone by modeling the emitters as a sparse vector in a location basis. A sympathetic reader would care because the resulting recovery algorithm then produces explicit expressions for the smallest separation between emitters that can be resolved and the weakest emitter power that remains detectable. These limits are stated as functions of measurement noise and sensor placement without any requirement for environmental mapping or emitter-specific signatures. The analysis is supported by both algorithmic simulations and a reconstruction performed on field-collected data.

Core claim

The recovery problem for multiple emitters using RSS data is formulated as one of sparse approximation or compressed sensing in a location basis. An appropriate recovery algorithm is investigated and used to characterize the limiting resolution and lowest detectable emitter power as functions of noise level, sensor geometry, and other variables. A reconstruction based on sampled data collected in the field illustrates the reasonableness of the parameter choices and conclusions.

What carries the argument

Sparse representation of emitter locations in a discretized location basis, recovered by a compressed-sensing algorithm applied to the RSS measurements.

If this is right

  • Resolution between emitters improves directly with reduced noise variance or with changes in sensor geometry that increase the effective rank of the measurement matrix.
  • The lowest detectable emitter power decreases as noise drops or as sensor placement improves the conditioning of the sparse recovery.
  • The number of emitters need not be known in advance; the algorithm recovers both the count and the positions.
  • No pre-collected RSS map of the environment is required for the recovery to proceed.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same sparse model could be used to predict how adding or moving sensors changes the achievable resolution before any new data are collected.
  • If the discretization grid is too coarse, the recovered positions will exhibit quantization bias whose size scales with grid spacing.
  • The detectability thresholds supply a quantitative baseline against which RSS performance can be compared to time-of-arrival or angle-of-arrival methods in the same geometry.

Load-bearing premise

The RSS values can be written as a sparse linear combination over a dictionary of possible emitter positions, and a standard recovery algorithm can invert that combination without additional side information.

What would settle it

An experiment that places two emitters closer than the predicted resolution distance at the measured noise level and checks whether the algorithm returns them as two distinct sources rather than one.

Figures

Figures reproduced from arXiv: 1907.01592 by Deborah Walter, Kurt Bryan.

Figure 2
Figure 2. Figure 2: Average recovered power from 500 simulation runs (white is 0 power, black is power 1 or higher), erroneous pathloss exponent. The true emitters are marked as stars, sensor locations as crosses. The lognormal randomized noise is simulated with σdB = 3. IV. ANALYSIS OF RESOLUTION AND DETECTION LIMITS The goal in this section is to develop a method for quantifying the local resolution one can obtain at any fi… view at source ↗
Figure 1
Figure 1. Figure 1: Average recovered power from 500 simulation runs (white is 0 power, black is power 1 or higher). The true emitters are marked as stars, sensor locations as crosses. The lognormal randomized noise is simulated with σdB = 3. recovery clustered around the three emitters, which are reasonably well resolved. The spread of each cluster gives an indication of the resolution one can achieve with this sensor config… view at source ↗
Figure 3
Figure 3. Figure 3: Simulated and approximated resolution probability for various [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Probability of successful resolution as function of [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Average recovered power from 500 simulation runs (white is 0 power, black is power 1 or higher). The true emitters are marked as stars, sensor locations as crosses. approach there are little improvements in resolution to be obtained by using too fine of a grid. C. Comparison to Cramer-Rao Bounds Other authors (e.g., [24], [25], [23]) have exam￾ined statistical bounds, for example, Cramer-Rao bounds, on the… view at source ↗
Figure 6
Figure 6. Figure 6: At power level 0.25 the emitter becomes essentially invisible [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: The normalized RSS measured from sensors randomly placed [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
read the original abstract

In this paper we investigate the problem of locating multiple non-cooperative radio frequency (RF) emitters using only received signal strength (RSS) data. We assume that the number of emitters is unknown and that individual emitters cannot be distinguished in the RSS data. Moreover, we assume that the environment in which the data has been collected has not been mapped or "fingerprinted" by the prior collection of RSS data. Our primary interest is the limiting resolution that can be obtained by this type of data, and the lowest power emitters that can be detected, as a function of noise level, sensor geometry, and other variables. We formulate the recovery problem as one of sparse approximation or compressed sensing, and investigate an appropriate recovery algorithm for this setting, and use it to illustrate our conclusions. We also include a reconstruction based on sampled data we collected, to illustrate the reasonableness of our parameter choices and conclusions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims that the limiting resolution and lowest detectable emitter power for multiple non-cooperative, indistinguishable RF emitters can be characterized as functions of noise level, sensor geometry, and related variables by formulating RSS measurements as a sparse approximation problem y = A x + noise (with A derived from a deterministic path-loss model on a location grid) and applying compressed-sensing recovery algorithms, illustrated via simulations and a real-data collection example without fingerprinting.

Significance. If the sparse recovery conditions hold, the work provides a model-based framework for deriving functional performance limits of RSS geolocation, which is significant for practical scenarios where environmental mapping is unavailable. The emphasis on unknown emitter count and lack of distinguishability, combined with both algorithmic investigation and real-data validation, strengthens the contribution if the derivations are reproducible.

major comments (2)
  1. [Model formulation and sensing matrix construction] The central claim that resolution and detectability limits can be characterized via sparse recovery rests on the sensing matrix A being exactly matched to the data-generating process (fixed path-loss exponent, on-grid emitters). The manuscript should explicitly address (e.g., via additional analysis or simulations in the recovery section) how performance degrades under log-normal shadowing or off-grid mismatch, as these violate the RIP/coherence conditions needed for reliable support recovery and directly impact the claimed functional dependence on noise and geometry.
  2. [Numerical results and real-data example] Table or figure presenting the numerical illustrations of resolution/detectability: if these results are generated only under exact model match (no shadowing, perfect grid alignment), they characterize performance only for the idealized case and do not fully support the general conclusions about limits in realistic environments.
minor comments (2)
  1. [Real-data reconstruction] The real-data example is helpful for assessing parameter reasonableness, but additional details on grid discretization, choice of path-loss exponent, and how the collected data aligns with the assumed model would improve clarity.
  2. [Abstract and introduction] Ensure the abstract and introduction consistently define key terms such as 'limiting resolution' and 'lowest detectable power' with reference to the sparse recovery metrics used later.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment below and indicate where revisions will be made to strengthen the presentation.

read point-by-point responses

  1. Referee: [Model formulation and sensing matrix construction] The central claim that resolution and detectability limits can be characterized via sparse recovery rests on the sensing matrix A being exactly matched to the data-generating process (fixed path-loss exponent, on-grid emitters). The manuscript should explicitly address (e.g., via additional analysis or simulations in the recovery section) how performance degrades under log-normal shadowing or off-grid mismatch, as these violate the RIP/coherence conditions needed for reliable support recovery and directly impact the claimed functional dependence on noise and geometry.

    Authors: The core contribution of the work is the derivation of resolution and detectability limits under the deterministic path-loss model with on-grid emitters, which permits direct use of sparse recovery guarantees. We agree that log-normal shadowing and off-grid mismatch represent important practical deviations that would degrade performance relative to the idealized bounds. The real-data example already incorporates such effects implicitly. In the revision we will add a dedicated paragraph in the recovery section with new simulations quantifying the degradation for moderate shadowing variance and small grid mismatches, while preserving the main functional characterizations for the matched case. revision: yes

  2. Referee: [Numerical results and real-data example] Table or figure presenting the numerical illustrations of resolution/detectability: if these results are generated only under exact model match (no shadowing, perfect grid alignment), they characterize performance only for the idealized case and do not fully support the general conclusions about limits in realistic environments.

    Authors: The numerical results are generated under exact model match to illustrate the theoretical limits obtained from the sparse-approximation analysis. The real-data collection example is presented precisely to demonstrate behavior outside the ideal model. We will revise the text surrounding the numerical figures and the real-data section to explicitly distinguish the idealized simulations from the practical validation, thereby clarifying the scope of the claimed limits. revision: partial

Circularity Check

0 steps flagged

No circularity: model-based dictionary and CS recovery are external to the claimed limits

full rationale

The abstract and provided text formulate RSS geolocation as y = A x + noise with A constructed from a deterministic path-loss model (fixed exponent, on-grid) and apply standard sparse recovery. No equations, fitted parameters, or self-citations are shown that reduce the claimed resolution/detectability limits to the inputs by construction. The derivation therefore rests on external CS theory and the modeling assumptions rather than self-referential steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract only; no explicit free parameters, axioms, or invented entities are stated.

pith-pipeline@v0.9.0 · 5685 in / 993 out tokens · 29686 ms · 2026-05-25T10:25:40.981380+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 40 canonical work pages

  1. [1]

    Network based wireless location,

    A. H. Sayed, A. Tarighat and N. Khajehnouri, “Network based wireless location,” IEEE Signal Processing Magazine, vol. 22, no. 4, pp 24-40, July 2005. doi:10.1109/MSP.2005.1458275

    work page doi:10.1109/msp.2005.1458275 2005

  2. [2]

    Sparsity-Aware Wireless Networks: Local- ization and Sensor Selection,

    Jamali-Rad, Hadi, “Sparsity-Aware Wireless Networks: Local- ization and Sensor Selection,” Ph.D. dissertation, Dept. Dept. of Microelectronics and Computer Engineering, TU Delft, 2014

    work page 2014

  3. [3]

    Received Signal Strength based indoor positioning based on compressed sensing,

    C. Feng, W.S. Anthea Au, Shahrokh Valaee, and Zhenhui Tan. “Received Signal Strength based indoor positioning based on compressed sensing,” IEEE Transactions on Mobile Com- puting, vol. 22, no. 12, pp. 1983-1993, December 2012. doi:10.1109/TMC.2011.216

    work page doi:10.1109/tmc.2011.216 1983

  4. [4]

    On the accuracy of a cellular location sys- tem based on RSS measurements

    A.J. Weiss, “On the accuracy of a cellular location sys- tem based on RSS measurements.” IEEE Trans on V e- hicular Technology , vol.52, no.6, pp. 1508-1518, 2003. doi:10.1109/TVT.2003.819613

    work page doi:10.1109/tvt.2003.819613 2003

  5. [5]

    Locating the nodes: coopera- tive localization in wireless sensor networks,

    N. Patwari, J.N. Ash, S. Kyperountas, A.O. Hero III, R.L. Moses, and N.S. Correal, “Locating the nodes: coopera- tive localization in wireless sensor networks,” IEEE Signal Processing Magazine , vol.22, no.4, pp. 54- 69, July 2005. doi:10.1109/MSP.2005.1458287

    work page doi:10.1109/msp.2005.1458287 2005

  6. [6]

    Compressed Sensing Technologies and Chal- lenges for Aerospace and Defense RF Source Localiza- tion,

    P. Daponte, L. De Vito, F. Picariello, S. Rapuano and I. Tudosa, “Compressed Sensing Technologies and Chal- lenges for Aerospace and Defense RF Source Localiza- tion,” 2018 5th IEEE International Workshop on Metrology for AeroSpace (MetroAeroSpace) , Rome, 2018, pp. 634-639. doi:10.1109/MetroAeroSpace.2018.8453560

    work page doi:10.1109/metroaerospace.2018.8453560 2018

  7. [7]

    Distributed spectrum sensing for cognitive radio networks by exploiting sparsity

    J.A. Bazerque, and G.B. Giannakis. “Distributed spectrum sensing for cognitive radio networks by exploiting sparsity.” IEEE Transactions on Signal Processing 58.3 (2010): 1847-

    work page 2010

  8. [8]

    doi:10.1109/TSP.2009.2038417

    work page doi:10.1109/tsp.2009.2038417 2009

  9. [9]

    Joint spectrum sensing and primary user localization for cognitive radio via compressed sensing

    Li, Xue, et al. “Joint spectrum sensing and primary user localization for cognitive radio via compressed sensing.” MILCOM 2010 MILITARY COMMUNICATIONS CONFERENCE, San Jose CA, November 2010. doi:10.1109/MILCOM.2010.5680334

    work page doi:10.1109/milcom.2010.5680334 2010

  10. [10]

    Low Cost, Low Complexity Sensor Design for Non-Cooperative Geolocation via Received Signal Strength,

    M.S. Butler, “Low Cost, Low Complexity Sensor Design for Non-Cooperative Geolocation via Received Signal Strength,” Masters Thesis, Department of Electrical and Computer Engi- neering, Air Force Institute of Technology, 2012

    work page 2012

  11. [11]

    Development of a Model and Localization Algorithm for Received Signal Strength-Based Geolocation,

    Amanda S. King, “Development of a Model and Localization Algorithm for Received Signal Strength-Based Geolocation,” Ph.D. Dissertation, Department of Electrical and Computer Engineering, Air Force Institute of Technology, 2013

    work page 2013

  12. [12]

    Multiple UA V tomography based geolocation of RF emitters,

    D.J. Walter, J. Klein, J. Kaupert, C. Bullmaster, and V . Chakravarthy, “Multiple UA V tomography based geolocation of RF emitters,” proceedings of Proc. SPIE 7707, Defense Transformation and Net-Centric Systems 2010 , April 5-9, 2010. doi:10.1117/12.850168

    work page doi:10.1117/12.850168 2010

  13. [13]

    Localization of RF Emitters using Com- pressed Sensing with Multiple Cooperative Sensors,

    D.J. Walter, K. Bryan, J. Stephens, C. Bullmaster, and V . Chakravarthy, “Localization of RF Emitters using Com- pressed Sensing with Multiple Cooperative Sensors,” pro- ceedings of NAECON 2012 , Dayton OH, July 25-27 2012. doi:10.1109/NAECON.2012.6531060

    work page doi:10.1109/naecon.2012.6531060 2012

  14. [14]

    An introduction to IEEE STD 802.15.4

    J.T. Adams. “An introduction to IEEE STD 802.15.4.” proceed- ings of 2006 IEEE Aerospace Conference, Big Sky MT, pp. 1-8,

    work page 2006

  15. [15]

    doi:10.1109/AERO.2006.1655947

    work page doi:10.1109/aero.2006.1655947 2006

  16. [16]

    A Practical Evalua- tion of Radio Signal Strength for Ranging-based Localization,

    K. Whitehouse, C. Karlof, and D. Culler, “A Practical Evalua- tion of Radio Signal Strength for Ranging-based Localization,” ACM SIGMOBILE Mobile Comput. Commun. Rev. , vol. 11, no. 1, pp. 41-52, 2007. doi:10.1145/1234822.1234829

    work page doi:10.1145/1234822.1234829 2007

  17. [17]

    Guarantee- ing Practical Con- vergence in Algorithms for Sensor and Source Localization,

    B. Fidan, S. Dasgupta, and B.D.O. Anderson, “Guarantee- ing Practical Con- vergence in Algorithms for Sensor and Source Localization,” IEEE Transactions on Signal Process- ing, V ol. 56, No. 9, pp. 4458 - 4469, September 2008. doi:10.1109/TSP.2008.924138

    work page doi:10.1109/tsp.2008.924138 2008

  18. [18]

    Decentralized task distribution among cooperative UA Vs in surveillance systems applications,

    E.P. de Freitas, T. Heimfarth, A.M. Ferreira, C.E. Pereira, F.R. Wagner, and T. Larsson, “Decentralized task distribution among cooperative UA Vs in surveillance systems applications,” Sev- enth International Conference on Wireless On-demand Network Systems and Services (WONS) , pp. 121-128, Feb. 3-5, 2010. doi:10.1109/WONS.2010.5437123

    work page doi:10.1109/wons.2010.5437123 2010

  19. [19]

    Distributed target localization via spatial sparsity,

    V . Cevher, M. F. Duarte, and R. G. Baraniuk, “Distributed target localization via spatial sparsity,” 2008 16th European Signal Processing Conference (EUSIPCO) 2008 , Lausanne, Switzerland, pp. 25-29 August 2008

    work page 2008

  20. [20]

    Multiple Target Localization Using Compressive Sensing,

    C. Feng, S. Valaee, Z. Tan, “Multiple Target Localization Using Compressive Sensing,” proceeding ofGLOBECOM 2009 - 2009 IEEE Global Telecommunications Conference , Honolulu HI, December 2009. doi:10.1109/GLOCOM.2009.5425808

    work page doi:10.1109/glocom.2009.5425808 2009

  21. [21]

    Compressed Remote Sensing of Sparse Objects,

    A.C. Fannjiang, T. Strohmer, and P. Yan, “Compressed Remote Sensing of Sparse Objects,” SIAM J. Imaging Sci. , 3(3), 595- 618, doi:10.1137/090757034

    work page doi:10.1137/090757034

  22. [22]

    Radio-Frequency Transmitter Geolocation Us- ing Non-Ideal Received Signal Strength Indicators,

    S. Whiting, “Radio-Frequency Transmitter Geolocation Us- ing Non-Ideal Received Signal Strength Indicators,” Masters Thesis, Department of Electrical and Computer Engineering, University of Utah, 2018

    work page 2018

  23. [23]

    Performance Analysis of RSS Fingerprinting Based In- door Localization,

    X. Tian, R. Shen, D. Liu, Y . Wen and X. Wang, “Performance Analysis of RSS Fingerprinting Based In- door Localization,” IEEE Transactions on Mobile Com- puting, vol. 16, no. 10, pp. 2847-2861, 1 Oct. 2017. doi:10.1109/TMC.2016.2645221

    work page doi:10.1109/tmc.2016.2645221 2017

  24. [24]

    Relative location estimation in wireless sensor networks,

    N. Patwari, A.O. Hero, M. Perkins, N.S. Correal, R.J. O’Dea. “Relative location estimation in wireless sensor networks,” IEEE Transactions on signal processing 51.8 (2003): 2137-

    work page 2003

  25. [25]

    doi:10.1109/TSP.2003.814469

    work page doi:10.1109/tsp.2003.814469 2003

  26. [26]

    Barankin bounds for position estimation using received signal strength measurements,

    H. Koorapaty, “Barankin bounds for position estimation using received signal strength measurements,” 2004 IEEE 59th V ehicular Technology Conference. VTC 2004-Spring (IEEE Cat. No.04CH37514), vol. 5, pp 2686-2690, 2004. doi:10.1109/VETECS.2004.1391408

    work page doi:10.1109/vetecs.2004.1391408 2004

  27. [27]

    Distance Estimation From Received Signal Strength Under Log-Normal Shadowing: Bias and Variance,

    S.D. Chitte, S. Dasgupta, Z. Ding, “Distance Estimation From Received Signal Strength Under Log-Normal Shadowing: Bias and Variance,” IEEE Signal Processing Letters , vol. 16, issue 3, March 2009. doi:10.1109/LSP.2008.2012229

    work page doi:10.1109/lsp.2008.2012229 2009

  28. [28]

    Source localization from received sig- nal strength under lognormal shadowing,

    Sree Divya Chitte, “Source localization from received sig- nal strength under lognormal shadowing,” Dept. of Electri- cal and Computer Engineering, University of Iowa, 2010. doi:10.17077/etd.getg4038

    work page doi:10.17077/etd.getg4038 2010

  29. [29]

    ESPRIT-estimation of signal parameters via rotational invariance techniques,

    R. Roy and T. Kailath, “ESPRIT-estimation of signal parameters via rotational invariance techniques,”IEEE Trans. On Acoustics, Speech, and Signal Processing , vol. 37, no. 7, pp. 984-995,

    work page

  30. [30]

    doi:10.1109/29.32276

    work page doi:10.1109/29.32276

  31. [31]

    Multiple emitter location and signal parameter estimation,

    R. Schmidt, “Multiple emitter location and signal parameter estimation,” IEEE Trans. On Antennas and Propagation , vol. 34, no. 3, pp 276-280, 1986. doi:10.1109/TAP.1986.1143830

    work page doi:10.1109/tap.1986.1143830 1986

  32. [32]

    Modeling and mitigating noise and nuisance 14 parameters in received signal strength positioning,

    R.K. Martin, A. King, J. Pennington, R. Thomas, R. Lenahan, and C. Lawyer, “Modeling and mitigating noise and nuisance 14 parameters in received signal strength positioning,” IEEE Trans. Signal Processing, vol. 60, no. 10, pp 5451-5463, October 2012. doi:10.1109/TSP.2012.2207118

    work page doi:10.1109/tsp.2012.2207118 2012

  33. [33]

    Foucart and H

    S. Foucart and H. Rauhut, A Mathematical Introduction to Com- pressive Sensing , New York: Springer Science and Business Media, 2013

    work page 2013

  34. [34]

    Stable recovery of sparse overcomplete representations in the presence of noise,

    D. Donoho, M. Elad, VN. Temlyakov, “Stable recovery of sparse overcomplete representations in the presence of noise,” IEEE Trans. Inform. Theory , vol. 52 pp. 6-18, Jan. 2006. doi:10.1109/TIT.2005.860430

    work page doi:10.1109/tit.2005.860430 2006

  35. [35]

    Sparse approximate solutions to linear systems,

    B.K. Natarajan, “Sparse approximate solutions to linear systems,”, SIAM J. Comput. , vol. 24, pp 227-234, 1995. doi:10.1137/S0097539792240406

    work page doi:10.1137/s0097539792240406 1995

  36. [36]

    Coherence pattern-guided com- pressive sensing with unresolved grids,

    A. Fannjiang and W. Liao, “Coherence pattern-guided com- pressive sensing with unresolved grids,” SIAM J. Imag- ing Sciences , vol. 5, no. 1, pp. 179-202, Feb. 2012. doi:10.1137/110838509

    work page doi:10.1137/110838509 2012

  37. [37]

    An Introduction To Compressive Sampling,

    E.J. Cand ´es, and M.B. Wakin,“An Introduction To Compressive Sampling,” IEEE SIGNAL PROCESSING MAGAZINE , vol. 21, March 2008. doi:10.1109/MSP.2007.914731

    work page doi:10.1109/msp.2007.914731 2008

  38. [38]

    Approximating a Sum of Random Variables with a Lognormal,

    N.B. Mehta and A.F. Molisch, “Approximating a Sum of Random Variables with a Lognormal,” IEEE Trans. of Wireless Communications, vol.6, No. 7, pp. 2690-2699, July 2007. doi:10.1109/TWC.2007.051000

    work page doi:10.1109/twc.2007.051000 2007

  39. [39]

    A low-cost desktop software defined radio design environment using MATLAB, simulink, and the RTL-SDR,

    R.W. Stewart et al. “A low-cost desktop software defined radio design environment using MATLAB, simulink, and the RTL-SDR,” IEEE Communications Magazine, vol. 53, issue 9, pp. 64-71, September 2015. doi:10.1109/MCOM.2015.7263347

    work page doi:10.1109/mcom.2015.7263347 2015

  40. [40]

    Best practice in RSS measurements and ranging,

    A. Zanella, “Best practice in RSS measurements and ranging, ” IEEE Communications Surveys and Tutorials vol. 18, issue 4, pp. 2662-2686, 2016. doi:10.1109/COMST.2016.2553452

    work page doi:10.1109/comst.2016.2553452 2016