Access in 4G LTE Cellular Systems ... an experimental framework for 5G cognitive radio access in .... realtime, including generation of random user data, pilots.
Experimental Testbed for 5G Cognitive Radio Access in 4G LTE Cellular Systems Martin Danneberg, Rohit Datta, Andreas Festag and Gerhard Fettweis Vodafone Chair Mobile Communications Systems, TU Dresden, Germany Email: {martin.danneberg, rohit.datta, andreas.festag, fettweis}@ifn.et.tu-dresden.de Abstract—Cognitive radio technology addresses the limited availability of wireless spectrum and inefficiency of spectrum usage. Cognitive Radio (CR) devices sense their environment, detect spatially unused spectrum and opportunistically access available spectrum without creating harmful interference to the incumbents. In cellular systems with licensed spectrum, the efficient utilization of the spectrum as well as the protection of primary users is equally important, which imposes opportunities and challenges for the application of CR. This paper introduces an experimental framework for 5G cognitive radio access in current a 4G LTE cellular systems. It can be used to study CR concepts in different scenarios, such as 4G to 5G system migrations, machine-type communications, device-to-device communications, and load balancing. Using our framework, selected measurement results are presented that compare Long Term Evolution (LTE) Orthogonal Frequency Division Multiplex (OFDM) with a candidate 5G waveform called Generalized Frequency Division Multiplexing (GFDM) and quantify the benefits of GFDM in CR scenarios. Keywords: 5G Waveform; Coexistance; Cognitive Radio; Experimental Testbed; LTE waveform.
I. I NTRODUCTION Wireless spectrum in industrial, scientific and medical band (ISM) and mobile cellular bands is increasingly getting congested and the demand for data traffic is growing continuously. A static spectrum management by authorities, where spectrum is licensed for a long-term duration over large geographical regions, cannot allocate sufficient spectrum that meets the demand. With CR technology [1], intelligent radio devices have the capability to sense their environment and opportunistically access available radio spectrum. For mobile cellular networks, such as Third Generation Partnership Project (3GPP) Long Term Evolution – Advanced (LTE-A), CR may complement techniques for carrier aggregation that mold together disparate bands for wider channels, but allow for higher spectrum utilization. In todays 4G mobile cellular networks, CR is considered to extend the LTE operation over TV white space (TVWS) [2], i.e. the frequency spectrum available through the replacement of analogue by digital TV. CR enables a non-interfering coexistence between DVB-T for digital TV as primary users and LTE as secondary users. In future 5G mobile cellular systems, CR has the potential to be applied in various scenarios, including (i) non-LTE standard compliant techniques to allow for a smooth migration from 4G to 5G systems, (ii) machine-type communication (MTC), such as from smart meters, complementary to primary human-centric voice and mobile data communications, leveraging the relaxed data rate
and latency demands of MTC, (iii) device-to-device (D2D) communications without disturbing the primary systems, and (iv) redirecting high data rate transmissions via secondary links for the purpose of load balancing. In all scenarios, the underlying LTE OFDM modulation scheme facilitates a robust and reliable transmission, but creates large side lobes and interference to adjacent incumbent transmissions. For future 5G systems, alternative waveforms, such as GFDM [3], [4] are considered. In the context of CR, the new waveform has better spectral properties and is therefore more suitable for secondary systems due to their smaller impact on the primary system. The multiband GFDM is a relatively new idea for designing a multicarrier PHY. GFDM is block based multicarrier transmission scheme derived from filter bank approach, where the transmit data of each block is distributed in time and frequency and each subcarrier is pulse-shaped with an adjustable pulse shaping filter. GFDM is well suited for CR, as the choice of pulse shaping filters makes the out-of-band leakage extremely small [5]. Compared to OFDM, which has rectangular pulse shaping, GFDM with a choice of transmit pulse shaping causes lesser interference to the adjacent incumbent frequency bands. This technique improves the Adjacent Channel Leakage Ratio (ACLR) of the GFDM system by around 20 dB. Compared to Filter Bank Multi Carrier (FBMC), which has no cyclic prefix, GFDM with its tailbiting cyclic prefix [6] addresses the synchronization issues that was problematic in FBMC [7], [8]. CR technology has been subject of research efforts for several years. Some of the existing work addresses CR-enabled LTE systems, such as LTE in TVWS, but so far mainly a theoretical study [9], [10] with only few experimental work. This paper introduces an experimental framework for CRenabled LTE, which can be used to study CR concepts in evolved 4G and future 5G systems. Based on the experimental testbed, the present paper studies the impact of a secondary CR-enabled system on the primary LTE OFDM and compares OFDM with GFDM as the secondary user in an MTC scenario. The remaining sections of the paper are structured as follows. Section II describes the OFDM and GFDM system model, Section III gives an overview of the experimental testbed and Section IV presents the measurement results. II. OFDM AND GFDM SYSTEM MODEL GFDM is a multicarrier system with flexible pulse shaping. In this section, the GFDM system model is described in detail as a basis for the experimimental work in the next section. In GFDM, at first, binary data is modulated and divided
dk [m]
d[]
binary data
N
dN k [n]
.. .
S/P
wkn
g[n]
xk [n]
.. .
x[n]
.. .
N
D/A
D
X
Fig. 1: GFDM transmitter system model
into sequences of KM complex valued data symbols. Each such sequence d[], = 0 . . . KM − 1, is spread across K subcarriers and M time slots for transmission. The data can be represented conveniently by means of a block structure ⎛
d0 d1 .. .
⎜ ⎜ D=⎜ ⎝
dK−1
⎞
⎛ d0 [0] ⎟ ⎟ ⎜ .. ⎟=⎝ . ⎠ dK−1 [0]
... ...
⎞ d0 [M − 1] ⎟ .. ⎠ , (1) . dK−1 [M − 1]
where dk [m] ∈ C is the data symbol transmitted on the k th subcarrier and in the mth time slot. A. Transmitter Model The GFDM transmitter structure is shown in Fig. 1. Consider the k th branch of the transmitter. The complex data symbols dk [m], m = 0, . . . , M − 1 are upsampled by a factor N , resulting in
Fig. 2: Measurement setup ⎛
dN k [n]
=
M −1
dk [m]δ[n − mN ],
n = 0, . . . , N M − 1 (2)
m=0
where δ[·] is the Dirac function. N Consequently, dN k [n = mN ] = dk [m] and dk [n = mN ] = 0. With filter length L ≤ M , the pulse shaping filter g[n], n = 0 . . . LN − 1, is applied to the sequence dN k [n]. Additional rate loss from filtering is avoided with tail biting technique described in [3], [11], followed by digital subcarrier upconversion. The resulting subcarrier transmit signal xk [n] can be mathematically expressed as kn xk [n] = (dN k g)[n] · w
(3) 2π
where denotes circular convolution and wkn = ej N kn . N is the upsampling factor that is necessary to pulse shape each of the subcarriers respectively and in this paper we consider N = K. Similar to (1), the transmit signals can also be expressed in a block structure
⎜ ⎜ X=⎜ ⎝
x0 x1 .. . xK−1
⎞
⎛ x0 [0] ⎟ ⎟ ⎜ .. ⎟=⎝ . ⎠ xK−1 [0]
... ...
⎞ x0 [M N − 1] ⎟ .. ⎠. . xK−1 [M N − 1]
(4) The transmit signal for a data block D is then obtained by summing up all subcarrier signals according to x[n] =
K−1
xk [n].
(5)
k=0
The result is then passed to the digital-to-analog converter and sent over the channel. According to the model OFDM can be seen as a special case of GFDM, where M = 1 and rectangular pulse shaping is applied. Cyclic prefixed single carrier (CP-SC) transmission is another special case, where K = 1 and there is no restriction to the filter. Hence, GFDM can be thought of as a generalized case of frequency division multiplexing, where OFDM and single-carrier transmission are the two particular modes of
TABLE I: System parameters (∗ valid only for GFDM) Parameters of the primary system.
transmission. The receiver model of GFDM is described in detail in [4].
FFT size Bandwidth Subcarrier spacing Modulation Uplink (UL) frequency Downlink (DL) frequency Used PRBs Waveform
III. E XPERIMENTAL FRAMEWORK FOR CR- ENABLED LTE The LTE/LTE-A testbed, developed by the Vodafone chair Mobile Communications Systems at the TU Dresden, is an experimental wireless testbed to study CR in cellular systems with a focus on physical layer aspects of LTE OFDM (4G) and future candidate waveforms for 5G cellular systems. The testbed is indoor and outdoor, and consists of up to three base stations and three user equipment devices. Two base stations reside at rooftop level. Its LTE-like cellular infrastructure, as described in [12], allows to measure relevant network parameters such as Bit Error Rate (BER) or outage events. The testbed is equipped with SIGNALION (a National Instruments company) test hardware. It basically consists of an FPGA powered baseband unit and a radio frontend unit. Depending on the firmware the equipment can either be configured as Evolved Node B (eNB) or user equipment (UE). Basic operating functionality is provided by hardware in realtime, including generation of random user data, pilots and control information at the transmitter side, as well as synchronization and decoding of the control information at the receiver side. The physical layer is based on LTE release 8 with some deviations, e.g. the uplink (UL) uses OFDMA instead of SC-FDMA. Relevant performance metrics of interest, such as RSSI, RSRP and SINR, are available in realtime. Further processing of measured data can then be done offline, typically using MATLAB. The IQ signals are logged either at the eNB or the UE. From the logged data, QAM constellations, CSI and BER can be calculated. Fig. 2 illustrates the testbed setup. The primary user downlink (DL) is connected from eNB to UE. The UE UL signal is added to the CR UL signal via a power combiner. This combined signal is afterwards divided into one signal for the eNB and one for the spectrum analyzer by a power splitter. The eNB receive port has a 20 dB attenuator to prevent clipping. In this work the power difference between the primary and secondary user is of more interest than the exact power levels.
2048 20 MHz 15 KHz QPSK 1.99 GHz 2.18 GHz 30 OFDM
Parameters of the secondary system. FFT size Bandwidth GFDM Blocksize Modulation Waveform Allocated subcarriers Roll-of-factor∗ Cylic prefix∗ Filter∗
512 20 MHz 15 QPSK OFDM or GFDM 100 0.1 With tail biting Raised Cosine
The UL signals are captured and the calculated BER is used as a link quality metric. IV. M EASUREMENT RESULT Fig. 3 depicts the different out of band leakage levels of OFDM and GFDM measured at the spectrum analyzer. GFDM performs up to 30 dB better than OFDM. This result exemplifies that GFDM is a promising candidate for CR coexistence between incumbent communication techniques. Furthermore, the impact from the secondary waveform on the primary user is analyzed. Fig. 4 illustrates the spectrum usage by the primary system: The area shown in green color is regarded as free. Therefore the secondary system will utilize it as is depicted in Fig. 5. Once the primary system is set up, the secondary system occupies a portion (≈ 4 MHz) of the available frequency resources. During the measurements, the position of the allocated subcarriers and the transmission power of the secondary system are altered in order to study the influence of the transmit power and size of guard bands on the primary system. −50
Power [dBm]
−60 −70 −80
Spectrum UE UL White space Control channels
−90 −100 1.98
Fig. 3: Out of band leakage of OFDM and GFDM
1.985 1.99 1.995 Frequency [GHz]
2
Fig. 4: Spectrum usage by primary user
−30 50
OFDM GFDM
Power difference [dB]
−40
Power [dBm]
−50 −60 −70 −80
9.6792 dB
40 30 20 10
−90
118.44 KHz −100 1.98
1.985 1.99 1.995 Frequency [GHz]
2
Fig. 5: The CR devices utilizes the free space in the LTE UL spectrum (marked in red color) with the GFDM waveform.
These measurements are carried out for OFDM and GFDM as the secondary waveform. Fig. 6 shows the power difference between the secondary and the primary system when the LTE UL frames are still decodeable. That means any further power increase of the secondary system will interrupt the primary communication. The Fig. 6 also depicts that GFDM has a lower influence on the primary system than OFDM and the power level of the CR device can be higher. The power ratio is about 9.7 dB. GFDM also needs about 118 kHz lesser spacing to a primary system as OFDM. However the effect is not as much as in Fig. 3. The reason is that the Automatic Gain Control (AGC) at the eNB works in time domain to ensure that the best possible dynamic range will be available from the analog-to-digital converter. The stronger 5G CR signal is also part of the received time signal and used as a reference by the AGC. Therefore the quantization of the primary signal is too bad and decoding is difficult. This could be a problem for consumer devices (e.g. laptop, smartphone), in particular when a CR shares the same antenna with an incompatible system or is placed close to it. As a result, the BER at the primary system is not only affected by the out of band leakage but also by the quantization noise. A solution could be to use tunable filters before the AGC at the primary system. V. C ONCLUSIONS The aim of this paper was to measure the impact of secondary 5G system transmission schemes on the performance of a primary 4G system. As a result it is shown that the coexistence of OFDM-based LTE and GFDM is possible. Based on the experimental work, it can be concluded that GFDM with lower out of band leakage – compared to OFDM – is more suitable as a next generation CR waveform to access frequency holes in an LTE cellular system.
0 1.98
1.985 1.99 1.995 Frequency of first subcarrier [GHz]
Fig. 6: Measured power difference between primary and secondary waveform, when LTE UL is still decodable at eNB. ACKNOWLEDGMENT This work has been performed in the framework of the ICT project ICT-258301 ”CREW” and ICT-318555 ”5GNOW” which are partly funded by the European Union. R EFERENCES [1] J. Mitola, Cognitive Radio: An Integrated Agent Architecture for Software Defined Radio, Ph.D. thesis, KTH Royal Institute of Technology, August 1999. [2] Electronic Communications Committee (ECC) within CEPT, “Measurements on the Performance of DVT Receivers in the Presence of Interference from the Mobile Service (Especially from LTE),” Report 148, ECC, June 2010. [3] G. Fettweis, M. Krondorf, and S. Bittner, “GFDM - Generalized Frequency Division Multiplexing,” in 69th IEEE VTC Spring, April 2009, pp. 1–4. [4] R. Datta, N. Michailow, M. Lentmaier, and G. Fettweis, “GFDM Interference Cancellation for Flexible Cognitive Radio PHY Design,” in IEEE VTC Fall, Quebec City, Canada, September 2012, IEEE. [5] R. Datta and G. Fettweis, “Improved ACLR by Cancellation Carrier Insertion in GFDM Based Cognitive Radios,” in IEEE VTC Spring, Seoul, South Korea, May 2014. [6] R. Datta, D. Panaitopol, and G. Fettweis, “Analysis of Cyclostationary GFDM Signal Properties in Flexible Cognitive Radio,” in IEEE International Symposium on Communications and Information Technologies (ISCIT), Goldcoast, Australia, October 2012, pp. 663–667. [7] R. Datta, G. Fettweis, Z. Koll´ar, and P. Horv´ath, “FBMC and GFDM Interference Cancellation Schemes for Flexible Digital Radio PHY Design,” in 14th IEEE Euromicro Conference on Digital System Design (DSD), August 2011, pp. 335–339. [8] T. Fusco, A. Petrella, and M. Tanda, “Sensitivity of multi-user filter-bank multicarrier systems to synchronization errors,” in Communications, Control and Signal Processing, 2008. ISCCSP 2008. 3rd International Symposium on, 2008, pp. 393–398. [9] R. Datta, N. Michailow, S. Krone, M. Lentmaier, and G. Fettweis, “Generalized Frequency Division Multiplexing in Cognitive Radio,” in 20th European Signal Processing Conference (EUSIPCO), Bucharest, Romania, 2012, pp. 2679–2683. [10] L.S. Cardoso, M. Kobayashi, Oyvind Ryan, and M. Debbah, “Vandermonde Frequency Division Multiplexing for Cognitive Radio,” in 9th IEEE Workshop on Signal Processing Advances in Wireless Communications (SPAWC), Recife, Brazil, July 2008, pp. 421–425. [11] N. Michailow, M. Lentmaier, P. Rost, and G. Fettweis, “Integration of a GFDM Secondary System in an OFDM Primary System,” in Future Network Summit, Warsaw, Poland, June 2011. [12] R. Irmer et al., “Multisite Field Trial for LTE and Advanced Concepts,” IEEE Communications Magazine, vol. 47, no. 2, pp. 92–98, 2009.