+.IEEE
`
`Advancing Technology
`for Humanity
`
`DECLARATION OF GERARD P. GRENIER
`
`I, Gerard P. Grenier, am over twenty-one (21) years of age. I have never been convicted
`of a felony, and I am fully competent to make this declaration. I declare the following to be true
`to the best of my knowledge, information and belief:
`
`1. I am Senior Director of Content Management of The Institute of Electrical and
`Electronics Engineers, Incorporated ("IEEE").
`
`2. IEEE is a neutral third party in this dispute.
`
`3. Neither I nor IEEE itself is being compensated for this declaration.
`
`\
`
`4. Among my responsibilities as Senior Director of Content Management, I act as a
`custodian of certain records for IEEE.
`
`5. I make this declaration based on my personal knowledge and information contained
`in the business records of IEEE.
`
`6. As part of its ordinary course of business, IEEE publishes and makes available
`technical articles and standards. These publications are made avai lable for public
`download through the IEEE digital library, IEEE Xplore.
`
`7. It is the regular practice of IEEE to publish articles and other writings including
`article abstracts and make them available to the public through IEEE Xplore. IEEE
`maintains copies of publications in the ordinary course of its regularly conducted
`activities.
`
`8. The article below has been attached as Exhibit A to this declaration:
`
`A. G. Bark, "Power control and active channel selection in an LPI FH system
`for HF communications", Proceedings of MILCOM 97, November 3-5,
`1997.
`
`9. I obtained a copy of Exhibit A through IEEE Xplore, where it is maintained in the
`ordinary course of IEEE's business. Exhibit A is a true and correct copy of the
`Exhibit, as it existed on or about June 24, 2019.
`
`IEEE
`10. The article and abstract from IEEE Xplore shows the date of publication.
`Xplore populates this information using the metadata associated with the publication.
`
`445 Hoes Lane Piscataway, NJ 08854
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0001
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`11. G. Bark, "Power control and active channel selection in an LPI FH system for HF
`communications" was published in the Proceedings of MILCOM 97. MILCOM 97
`was held from November 3-5, 1997. Copies of the conference proceedings were
`made available no later than the last day of the conference. The article is currently
`available for public download from the IEEE digital library, IEEE Xplore.
`
`12. I hereby declare that all statements made herein of my own knowledge are true and
`that all statements made on information and belief are believed to be true, and further
`that these statements were made with the knowledge that willful false statements and
`the like are punishable by fine or imprisonment, or both, under 18 U.S.C. § 100 l.
`
`ar true~ c~ ~
`I declare under pen~Ity.ofpe~jury that the foregoing state22L1ents
`/ : ' ~
`Executed on: dS ·J(Jn~ ;).p) y
`
`_
`---"'-=-+---_.c;..-- - - - -
`
`j
`
`\
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0002
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`
`
`
`
`
`
`EXHIBIT A
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0003
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`Power control and active channel selection in an LPI FH system for HF communications - IEEE Conference Publication
`6/24/2019
` IEEE.org IEEE Xplore Digital L brary
`
`IEEE-SA
`
`IEEE Spectrum
`
`More Sites
` Cart (0) Create Account
`
`
` Personal Sign In
`
`|
`|
`|
`|
`|
`|
`
` Back to Results
`
`More Like This
`
`Power control in packet switched time
`division duplex direct sequence spread
`spectrum communications
`[1992 Proceedings] Vehicular Technology
`Society 42nd VTS Conference - Frontiers
`of Technology
`Published: 1992
`
`Low-complexity erasure insertion in
`frequency-hopping spread-spectrum
`communications subjected to fading and
`partial-band interference
`GLOBECOM'01. EEE Global
`Telecommunications Conference (Cat.
`No.01CH37270)
`Published: 2001
`
`View More
`
`See the top organizations
`patenting in technologies
`mentioned in this article
`
`Click to Expand
`
`Access provided by:
`IEEE Publications Operations
`Staff
`Sign Out
`
`
`
`Browse
`
`My Settings
`
`Get Help
`
`Conferences > MILCOM 97 MILCOM 97 Proceedings
`
`Power control and active channel selection in an LPI FH system
`for HF communications
`Publisher: IEEE
`
`Alerts
`
`Manage
`Content Alerts
`
`Add to Citation
`Alerts
`
` << Results
`
`1 Author(s)
`
`G. Bark View All Authors
`
`89
`Full
`Text Views
`
`2P
`
`atent
`Citations
`
`Abstract
`
`Authors
`
`References
`
`Citations
`
`Keywords
`
`Metrics
`
`More Like This
`
`Downl
`PDF
`
`Abstract: To improve the performance of frequency-hopping systems on interference-
`limited HF channels, so-called adaptive frequency-hopping (AFH), which uses an
`adaptively selected... View more
`
` Metadata
`Abstract:
`To improve the performance of frequency-hopping systems on interference-limited HF
`channels, so-called adaptive frequency-hopping (AFH), which uses an adaptively
`selected pool of the "best" hopping-frequencies for communication, has been proposed.
`We extend the adaptivity of the AFH scheme by adjusting the transmitted power on
`each channel individually and by adaptively changing the number N/sub a/ of active
`channels that are selected to the pool. Fewer active channels (up to a certain point) give
`improved communication performance since the used channels, on the average, are
`less interfered. However, by decreasing N/sub a/, the protection against hostile
`detection is decreased. This trade-off between communication and LPI (low probability
`of intercept) performance with respect to N/sub a/ is shown. Our analysis shows that the
`codeword error rate is minimized when about 20% of the channels are selected to the
`active pool, and that the LPI protection against the two tested hostile detectors, as
`expected, improves for larger N/sub a/. Generally, the hostile detectors require less
`received signal-to-interference ratio than the legal AFH receiver to obtain acceptable
`performance. For the parameters we have chosen in our duel simulation, the results
`indicate that the LPI performance seems to be more sensitive to the choice of active
`channel pool size than the communication performance.
`
`Published in: MILCOM 97 MILCOM 97 Proceedings
`
`https://ieeexplore.ieee.org/document/646773
`
`1/2
`
` Export to
`
`Collabratec
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0004
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`6/24/2019
`
`Power control and active channel selection in an LPI FH system for HF communications - IEEE Conference Publication
`Date of Conference: 3-5 Nov. 1997
`INSPEC Accession Number: 5958959
`
`Date Added to IEEE Xplore: 06 August
`2002
`
`Print ISBN: 0-7803-4249-6
`
`DOI: 10.1109/MILCOM.1997.646773
`
`Publisher: IEEE
`
`Conference Location: Monterey, CA, USA,
`USA
`
`Authors
`
`References
`
`Citations
`
`Keywords
`
`Metrics
`
`IEEE Account
`
`Profile Information
`
`Purchase Details
`
`Need Help?
`
`Other
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.
`© Copyright 2019 IEEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.
`US & Canada: +1 800 678 4333
`Worldwide: +1 732 981 0060
`
`IEEE Account
`
` Purchase Details
`
` Profile Information
`
` Need Help?
`
`» Change Username/Password
`» Update Address
`
`» Payment Options
`» Order History
`» View Purchased Documents
`
`» Communications Preferences
`» Profession and Education
`» Technical Interests
`
`» US & Canada: +1 800 678 4333
`» Worldwide: +1 732 981 0060
`» Contact & Support
`
`
` About EEE Xplore Contact Us
`|
`
`
`|
`
`Help
`
`
`|
`
`Accessibility
`
`
`|
`
`Terms of Use
`
`
`|
`
`Nondiscrimination Policy
`
`
`|
`
`Sitemap
`
`
`|
`
`Privacy & Opting Out of Cookies
`
`A not-for-profit organization, IEEE is the world's largest technical professional organization dedicated to advancing technology for the benefit of humanity.
`© Copyright 2019 EEE - All rights reserved. Use of this web site signifies your agreement to the terms and conditions.
`
`https://ieeexplore.ieee.org/document/646773
`
`2/2
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0005
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`POWER CONTROL AND ACTIVE CHANNEL SELECTION IN AN
`LPI FH SYSTEM FOR HF COMMUNICATIONS*
`
`Gunnar Bark
`Radio Communication Systems, Dept. of Signals, Sensors & Systems
`Royal Institute of Technology, Sweden, E-mail: gunbar@lin.foa.se
`
`ABSTRACT
`
`To improve the performance of frequency-hopping sys(cid:173)
`tems on interference-limited HF channels, so-called adap(cid:173)
`tive frequency-hopping (AFH), which uses an adaptively se(cid:173)
`lected pool of the "best" hopping-frequencies for the com(cid:173)
`munication, has been proposed. In this paper, we extend
`the adaptivity of the AFH scheme by adjusting the trans(cid:173)
`mitted power on each channel individually and by adap(cid:173)
`tively changing the number Na. of active channels that are
`selected to the pool. Fewer active channels (up to a certain
`point) give improved communication performance since the
`used channels, on the average, are less interfered. However,
`by decreasing Na., the protection against hostile detection is
`decreased. This trade-off between communication and LPI
`performance with respect to Na. is shown in the paper.
`
`Our analysis shows that the codeword error rate is mini(cid:173)
`mized when about 20 % of the channels are selected to the
`active pool, and that the LPI protection against the two
`tested hostile detectors, as expected, improves for larger Na..
`Generally, the hostile detectors require less received signal(cid:173)
`to-interference ratio than the legal AFH receiver to obtain
`acceptable performance. For the parameters we have chosen
`in our duel simulation, the results indicate that the LPI per(cid:173)
`formance seems to be more sensitive to the choice of active
`channel pool size than the communication performance.
`
`INTRODUCTION
`
`In military HF radio communications, there is an increasing
`interest in using spread-spectrum (SS) techniques, mainly
`due to the jamming protection and the low probability of
`intercept (LPI) properties connected with SS technology, [1].
`A main problem with HF communication is the severe inter(cid:173)
`ference from other users of the spectrum and this has to be
`considered when evaluating SS systems on the HF channel.
`Extensive measurements and modeling of the occupancy of
`the HF spectrum in Europe, see e.g. [2], [3], have been per(cid:173)
`formed and are the basis of the HF interference model that
`is used in this paper.
`
`*This work has been supported by the Defence Research Establish(cid:173)
`ment and the Defence Materiel Administration in Sweden.
`
`Up till now, frequency-hopping (FH) has been the dominat(cid:173)
`ing SS method in operational and test HF radio systems.
`A special form of frequency-hopping, that is treated in this
`paper, is adaptive FH that somehow tries to find a pool of
`the "best" hopping-frequencies, or channels, and use only
`these for transmission. Since the power of the interfering
`HF signals is very varying for different frequencies, one can,
`by using only the ·least interfered so-called active channels,
`gain a lot in communication performance compared to con(cid:173)
`ventional FH, [4], [5]. Since the AFH signal can "hide" in
`the dense HF interference environment, it also has good LPI
`properties, [6].
`
`In this paper, we generalize the AFH scheme in two ways.
`Firstly, the transmitted power on each active channel is
`adjusted so that all channels obtain the same signal-to(cid:173)
`interference ratio (SIR) at the receiver, which gives good
`LPI properties. Secondly, the number Na. of active channels
`is adaptively changed depending on the interference envi(cid:173)
`ronment. We will show how the communication and LPI
`performance are affected by the size of the pool of active
`channels. The LPI performance is evaluated with respect
`to two different hostile detectors; the basic wideband ra(cid:173)
`diometer and a more sophisticated matched filter detector
`with interference suppression capability.
`
`THE AFH SYSTEM AND MODELS
`
`The structure of the AFH scheme is shown in figure 1. It is
`based on a conventional FH system with narrowband digital
`modulation; for simplicity, we have chosen binary FSK with
`noncoherent detection. However, in the AFH system, the
`hopping is performed between the Na. active channels, which
`is a subset A of the total set of N channels. The PN gener(cid:173)
`ator in the transmitter therefore, for each hop, generates a
`pseudo-random symbol from an alphabet of size Na.. Each
`symbol is mapped to one of the active channels by a fre- ·
`quency map (a table). The channel number is then fed
`to the frequency-hopper, which transmits one channel sym(cid:173)
`bol on that particular channel. In the receiver, an identi(cid:173)
`cal PN generator and mapping function generate the same
`channel number to the de-hopper, so that an estimate m of
`the message m can be obtained after decoding.
`
`0-7803-4249-6/97/$10.00 © 1997 IEEE
`
`1031
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0006
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`-
`I
`
`-
`I
`
`I
`I HF
`(hannell
`A,S
`
`L -
`
`J
`
`PN
`generator
`
`PN
`generator
`
`Figure 1: The structure of the AFH system.
`
`To evaluate all channels in the system, there is a link quality
`analyzer (LQA) in the receiver. The LQA monitors all N
`channels, measures their interference power levels and cre(cid:173)
`ates the pool A of the Na active channels with the lowest in(cid:173)
`terference levels. Furthermore, the LQA assigns a required
`transmitting power level Si, i = 1, ... , Na, for each active
`channel according to the chosen power control scheme. The
`message to be transmitted is divided into frames of M bits.
`After a frame has been received, the LQA constructs a new
`pool A of active channels and a new vector S = S1 , ... , SN.
`of power levels. A and S are sent to the transmitter via a
`feed-back channel so that the frequency-hopping takes place
`among these Na active channels until the next updating of
`A and S is carried out.
`
`Since the interfering HF signals are rather stationary over
`time, [7], most of the active channels will stay good dur(cid:173)
`ing the whole frame length TF, which is set to about 10 s.
`Nevertheless, a number of active channels could become in(cid:173)
`terfered during TF which, on the average, gives an increas(cid:173)
`ing channel error rate during the frame. Therefore, error(cid:173)
`correcting codes with error/ erasure decoding is used to get
`acceptable performance. If an active channel becomes in(cid:173)
`terfered, the LQA observes that the SIR for that channel
`goes below a threshold ')'e and erases the remaining channel
`symbols in the frame that were transmitted on that chan(cid:173)
`nel. To spread the erased symbols over many codewords,
`interleaving over the whole frame TF is applied. In this
`investigation, we will not consider the effects of errors on
`the return channel. This restriction can be justified by the
`fact that the feed-back channel could be well protected by
`coding since it does not need to have a very large capacity.
`Furthermore, there are many military scenarios where the
`transmission on the return channel can use a lot more sig(cid:173)
`nal power than the forward channel; for example, a coast
`station transmitting to a submarine, where the link to the
`submarine is the return channel.
`
`The received AFH signal is assumed to be slow and flat
`Rayleigh fading compared to the bit duration and the mo-
`
`mentary bandwidth, respectively, of the AFH signal. The
`HF interference environment is dominated by narrowband
`interference from other HF users and is modeled according
`to [5]. On each FH channel, the interference is modeled
`as a stochastic Gaussian process, J( i) ( t) with variance, or
`power, Ii that is derived from the Laycock-Gott congestion
`model, [2]. The interference processes on all channels are
`modeled to change power levels independently of each other
`according to a Poisson process with, on the average, 2 min(cid:173)
`utes between each transition, [7].
`
`POWER CONTROL
`
`The task _of the power control scheme is to divide the lim(cid:173)
`ited resource of transmitter power between the active FH
`channels in the "best" way. Since the power control is rel(cid:173)
`atively slow (only one updating per each 10 s long frame),
`we do not try to counteract the multipath fading with it.
`It is instead intended to adapt to the fairly slowly varying
`HF interference from other users of the band. The inter(cid:173)
`ference power Ii and the (average) transmission loss Li on
`each channel are measured in the receiver and are used as
`input to the algorithm. Thus, the problem can be formu(cid:173)
`lated in its most condensed form as: Given I = ! 1 , ... , !Na,
`L = L1, ... , L N", and an average transmitted power S, how
`shall S be distributed between elements of S? Note that
`the restriction is on the average transmitted power and not
`on the peak power. In order to maximize the LPI perfor(cid:173)
`mance of the AFH system, we have chosen to adjust the
`transmitted power levels so that the received SIR, initially
`in each frame, will be the same on all active channels, i.e.,
`~ _§_
`SN.
`L1I1 - L2h - ... - LNJNa.
`
`(1)
`
`The transmitted power levels for this scheme are
`s
`
`Si=
`
`I:S:i:S:Na
`
`LJi="faLJi,
`N
`N
`1
`1
`a
`a
`NLLk·NLh
`a k=l
`where ,a is the SIR at the AFH receiver on all active chan(cid:173)
`
`a k=l
`
`(2)
`
`nels. We call this algorithm SISIR (Same Initial SIR). It
`has good LPI performance since the transmitted power lev(cid:173)
`els are not higher than what is needed to ensure a certain
`minimum SIR ('Ya) at the AFH receiver. Obviously, the
`choice of the threshold 'Ya is very important since it directly
`affects the number of active channels. For a fixed value of
`
`the average transmitted power S, a higher value of ,a re(cid:173)
`
`sults in fewer channels for which the SIR can be kept over
`'Ya· And for a lower ,a there are many channels that qualify
`for the pool A. Note also that since the interference envi(cid:173)
`ronment is (slowly) changing, the number of active channels
`will be different for every frame. Hence, the system auto(cid:173)
`matically adapts to the varying HF interference by changing
`the size of the active pool A.
`
`1032
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0007
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`We have also evaluated an algorithm that we call MINIBEP
`which aims at good communication performance by min(cid:173)
`imizing the initial bit error probability for every data
`frame, [8J. However, our analysis showed that MINIBEP
`and SISIR give approximately equivalent communication
`performance. The resulting bit errors are mostly due to fast
`changes of the interfering signals which we anyway cannot
`counteract with our relatively slow power control. Those
`errors are better mitigated with the error-correcting coding
`with erasure decoding. Furthermore, the simulation results
`demonstrated the superior LPI capability of the SISIR al(cid:173)
`gorithm which we have therefore used in this paper.
`
`COMMUNICATION PERFORMANCE
`
`(3)
`
`In this section, we present simulation results of the com(cid:173)
`munication performance for the AFH system with different
`values of 'Ya, and hence, different sizes of the pool A of
`active channels. The performance measure we use is the
`codeword error rate (CWER) after decoding. For the eval(cid:173)
`uations we have applied a BCH block code with block size
`n = 127, number of information symbols k = 64 and mini(cid:173)
`mum Hamming distance dmin = 21 and as mentioned above,
`interleaving over the whole frame is applied. The threshold
`'Ye, which decides if a received symbol should be erased, is
`set equal to 'Ya since it is reasonable to erase the remaining
`symbols on an active channel if the SIR on that channel goes
`below 'Ya during a frame. Let e and p denote the number
`of errors and erased bits, respectively, in a codeword. The
`decoder can correctly decode a codeword if
`2e + p + 1 :$ dmin ,
`otherwise a codeword error is assumed to occur. Further(cid:173)
`more, the total number of channels N is 256, the bit rate is
`500 bit/s and one channel symbol (bit) is transmitted per
`hop. The size of each frame is 80 codewords, which gives
`Tp = 10.24 s and 5120 information bits per frame. As men(cid:173)
`tioned above, the interference on each channel is modeled
`according to the Laycock-Gott congestion model and is as(cid:173)
`sumed to change average power according to a Poisson pro(cid:173)
`cess with, on the average, 2 minutes between each change.
`The transmission losses Li are set to be the same on all
`FH channels. In figure 2, we see simulated codeword error
`rates for different values of 'Ya· The CWER is plotted as a
`function of the expected value of SIR at the receiver on a
`channel arbitrarily chosen among all N channels. Since the
`interference powers on different channels are assumed to be
`randomly distributed according to the interference model,
`and the signal power varies due to fading, the SIR for an
`arbitrary channel is a stochastic variable with an expected
`value E[SIRJ. (Observe that this is not the E[SIR] on only
`the active channels where the actual communication takes
`place. That E[SIR] is higher).
`The simulation results show that the CWER decreases for
`
`10° c-----,----,----.----~---,---~
`
`··············· ......... : ................ ; .............. .
`
`. ·. \<< .. ~.
`
`········:···············
`
`······:······"-· ···•:,:····· ··········:·············
`'
`··: .....
`:
`:
`:
`'
`:··.'·, Y=t8dB
`.
`\ : ·· ... · ... ,!. :
`: Y.=26dB\
`'"-,
`:
`"
`·r,-30:dB ·,.,
`......................... : .. \~. ~ . .::f ........... ·,>.,:.,: ............ .
`... .
`'
`:•.
`. ·.
`'
`·22.dB
`'
`'
`.
`':
`~ .,
`10"' .___ _ _ _,_ _ _ __._ _ _ _ ,_· _,_._ __ _,_ _ _ __._ _ _ ____,
`-20
`-10
`0
`10
`20
`30
`40
`E [SIR) at the AFH receiver, (dB)
`
`10-3
`
`:
`
`Figure 2: The codeword error rate for the AFH system
`for different values of the threshold 'Ya· Dash-dotted curve:
`,a=18 dB, solid curve: ,a=22 dB, dashed curve: ,a=26 dB,
`dotted curve: ,a=30 dB.
`
`increasing threshold 'Ya up to a certain limit. When 'Ya is
`increased, the channels that are selected to the active pool
`A have a higher SIR which gives fewer channel errors and
`lower CWER. But if 'Ya is set too high, which gives only a
`few active channels, the CWER gets bigger since the possi(cid:173)
`ble changes of the interference on these few channels cause
`more erased symbols than the decoder can cope with. Our
`simulations have shown that for a E[SIRJ = 10 dB the av(cid:173)
`erage number Na of active channels are 138, 109, 82 and 60
`for 'Ya=18, 22, 26, 30 dB, respectively. From figure 2, we
`see that setting 'Ya=26 dB, results in best communication
`performance for our choice of simulation parameters. For
`a CWER of 10-3, 'Ya = 26 dB gives Na = 48, i.e., that
`about a fifth of the channels are selected to the active pool.
`However, by decreasing the number of used channels, the
`protection against hostile detection is decreased if the in(cid:173)
`terceptor can estimate which hopping-frequencies are used.
`This is investigated in the next section.
`
`LPI PERFORMANCE
`
`To evaluate the LPI performance of a radio system, one has
`to define the detector device that the hostile interceptor uses
`and that the system is exposed to. The choice of detection
`technique depends upon what the interceptor is supposed
`to know about the signal of interest and the characteristics
`of the radio environment. When nothing is known about
`the signal, the interceptor is forced to use a general-purpose
`detector, for instance, the radiometer that simply measures
`the received energy in a given time and frequency interval.
`The other extreme is that the interceptor knows everything,
`
`1033
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0008
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`0.9
`
`0.8
`
`§
`
`)o.s
`
`"C
`0
`.o0.5
`e a.
`
`-30
`
`-20
`
`Figure 3: The LPI performance of the AFH system for different
`ra when the interceptor uses the radiometer.
`
`Figure 4: The LPI performance of the AFH system for different
`'Ya when the interceptor uses the MF detector.
`
`except for the secret "key" of the signal, which in the AFH
`system is the sequence of hopping-frequencies. We have c~o(cid:173)
`sen to evaluate the LPI properties of the AFH system with
`respect to two detectors representing these extremes; the
`radiometer and the more sophisticated matched filter (MF)
`detector which has total knowledge of the AFH system and
`performs a maximum-likelihood suppression of the interfer(cid:173)
`ing signals [6]. Their performances give approximate bounds
`for the LPI capability of the AFH system with the radiome(cid:173)
`ter's detection performance as the best case from the com(cid:173)
`municator's point of view. As the AFH receiver, both de(cid:173)
`tectors are assumed to measure the interference power on
`each channel and use that in the detection process. The LPI
`performance measure we use is the probability Pv that the
`hostile detector detects the presence of the AFH signal.
`
`The LPI performance of the AFH system depends a lot on
`how well the interceptor can estimate which channels that
`are actually used [6]. Measurements have shown that the in(cid:173)
`terfering powers in the HF spectrum are strongly correlated
`for distances up to 500-1000 km, [9]. Therefore, we will in
`this paper investigate the case when the interfering powers
`at the sites of the AFH receiver and the interceptor are fully
`correlated. This is when the interceptor is relatively close to
`the receiver and is the worst case from the communicator's
`point of view. Hence, the interceptor can estimate which
`channels that are selected to the active pool A and only has
`to inspect those Na channels since the communication takes
`place on them. We have also made the, for the communica(cid:173)
`tor, pessimistic assumption that the interceptor estimates
`the pool A perfectly.
`
`Since we have not been able to obtain analytical perfor(cid:173)
`mance results for the two detectors, we have used simulation
`
`studies, and as before, N = 256, Rb= 500 bit/sand the hop
`rate is about 992 hop/s (500 · 127 /64). The detectors are
`allowed to measure the received signal during 0.1 second
`before they make a decision if the AFH signal is present
`or not and their false alarm probabilities , PFA, are set
`to 10%. In figure 3, we see some curves of the probabilities
`of detection, Po, as a function of E[SIR] on an arbitrary
`channel at the interceptor if the radiometer is used. The
`dashed, solid and dash-dotted curve is for ra=26, 22 and
`18 dB, respectively. We can notice that the Pv does not go
`below PFA, which is quite expected. The worst the detec(cid:173)
`tors can do, is to give an equally low alarm rate when the
`AFH signal is present, as when it is absent. As expected,
`the radiometer requires higher signal-to-interference ratio
`for the detection when 'Ya is decreased (Na is increased).
`For a larger number of active channels, there are simply
`more hopping-frequencies that the hostile detector has to
`monitor, which increases the total received interference and
`the uncertainty in the detection process.
`
`In figure 4, we see the detection performance of the more so(cid:173)
`phisticated MF detector (notice the different scale on the x(cid:173)
`axis). Compared to the radiometer, the MF detector proves
`to be a more serious threat to the AFH system due to the
`interference suppression in each channel in the MF detec(cid:173)
`tor, [6]. This function decreases the influence of the chan(cid:173)
`nels with high interference on the overall decision. In the
`radiometer, on the other hand, there is no such interference.
`suppression feature, and hence, the detected energy on all
`channels are equally important in its decision. Therefore,
`the radiometer has large difficulties in distinguishing the
`interference from the AFH signal when there are channels
`with strong interference, which deteriorates the radiometer's
`performance substantially. Furthermore, we see in figure 4
`
`1034
`
`Marvell Semiconductor, Inc. - Ex. 1014, Page 0009
`IPR2019-01350 (Marvell Semiconductor, Inc. v. Uniloc 2017 LLC)
`
`

`

`that the LPI performance of the AFH system depends a lot
`on the threshold 'Ya in the case of the MF detector as well.
`The difference in detection performance between the three
`values of 'Ya is about the same as for the radiometer.
`
`By combining the evaluations of the communication perfor(cid:173)
`mance and the LPI performance, we can now illustrate how
`the duel between the communicator and the hostile detec(cid:173)
`tor is affected by the choice of "fa. For example, if the AFH
`system changes 'Ya from 26 to 22 dB the interceptor requires
`about 8 dB in increased E[SIR] to still achieve PD = 0.5.
`If we compare that with the communication performance in
`figure 2, we see that the communicator loses only 3 dB when
`changing 'Ya from 26 to 22 dB for a codeword error rate of
`10-3 • Hence, if there is 3 dB more of transmitter power
`available, these results indicate that it is worthwhile for the
`communicator to use them for increasing the pool of active
`channels by decreasing 'Ya· Actually, for a required CWER
`of 10-3, the results show that it is beneficial to decrease 'Ya
`to 18 dB, which gives Na.= 206, i.e., about 80 % of all chan(cid:173)
`nels are then used for the active pool. Our simulations have
`shown that it is not worthwhile to decrease 'Ya even more.
`The penalty in communication performance then becomes
`bigger than the gain in LPI performance. We can also see
`that if the communicator requires CWER = 10-4 , there is
`no gain in decreasing 'Ya to 18 dB.
`
`The main conclusion of the results above is perhaps that the
`LPI performance seems to be more sensitive to the choice
`of 'Ya than the communication performance. However, it
`should be stressed again that the detection results in figure
`3 and 4 are under, for the communicator, pessimistic as(cid:173)
`sumption that the interceptor can estimate the number of
`active channels perfectly. If he could not do that, and in(cid:173)
`stead would detect on a fixed number of the least interfered
`channels, independently of the size of Na., the detection per(cid:173)
`formance would not vary so much for different values of 'Ya·
`
`CONCLUSIONS
`
`We have in this paper evaluated both the communication
`and LPI performance of an adaptive frequency-hopping sys(cid:173)
`tem. Two adaption techniques were introduced to increase
`the performance on the interference-limited HF channel. To
`obtain good LPI performance the transmitted power was
`accommodated so that the received SIR will be equal to a
`threshold 'Ya on all active channels. Additionally, to adapt
`to the varying HF environment, the number of active chan(cid:173)
`nels was adjusted for each data frame so that a SIR of 'Ya
`could be maintained on all channels that are used for the
`communication.
`
`As indicated by the results above, the setting of the thresh(cid:173)
`old 'Ya., which decides the number of active channels that
`are used for the communication, has great impact on the
`performance of the AFH system. The simulation results
`
`showed that the codeword error rate was minimized when
`about 20 % of the channels were selected to the active pool.
`However, the LPI protection against both the radiometer
`and the MF detector was shown to increase when the pool
`of active channels is made larger (lower "fa). The results
`showed that, generally, the legal AFH receiver requires a
`much higher SIR than the hostile detectors to get accept(cid:173)
`able performance. By combining the results from the com(cid:173)
`munication and LPI performance evaluations, we saw that
`selecting as many as about 80 % of the channels to the ac(cid:173)
`tive pool minimized the difference between the SIR that the
`hostile detectors require to achieve a probability of detection
`of 0.5, and the SIR that the AFH system needs to obtain
`a codeword error rate of 10-3 • However, this is provided
`that there is enough transmitter power available so that the
`SIR can be kept at the constant level 'Ya even on the most
`interfered active channels.
`
`References
`
`[1] M. K. Simon et al., Spread spectrum communications
`handbook, McGraw-Hill Inc., 1994.
`[2] P. J. Laycock, G. F. Gott et al., "A model for HF spec(cid:173)
`tral occupancy," Conj. Proc. HF radio systems a