PaperlayerView exact revision on arXiv

Source-linked AI summary

srsLTE: An Open-Source Platform for LTE Evolution and Experimentation

Ismael Gomez-Miguelez, Andres Garcia-Saavedra, Paul D. Sutton, Pablo Serrano, Cristina Cano, Douglas J. Leith

arXiv:1602.04629v1cs.NI

TL;DR

Existing LTE testbeds can be difficult to customize and instrument, motivating an open-source alternative. The paper presents a modular, Release 8-compliant srsLTE/srsUE platform and evaluates it through unlicensed LTE/WiFi coexistence experiments, finding hardware-dependent WiFi behavior.

  • Problem

    Existing LTE testbeds are limited in functionality or extensibility, or too complex to modify and customize.

  • Method

    The paper presents a modular, open-source LTE library and software-radio UE, using real hardware to support LTE extension, instrumentation, and coexistence experimentation.

  • Results

    WiFi coexistence results vary with LTE duty cycle, spectrum, transmit power, modulation and coding, and WiFi-card vendor.

  • Takeaways & Limitations

    Off-the-shelf wireless-card particularities may affect fair coexistence in ways difficult to predict through analysis or simulations.

Abstract

from arXiv · show

Testbeds are essential for experimental evaluation as well as for product development. In the context of LTE networks, existing testbed platforms are limited either in functionality and/or extensibility or are too complex to modify and customise. In this work we present srsLTE, an open-source platform for LTE experimentation designed for maximum modularity and code reuse and fully compliant with LTE Release 8. We show the potential of the srsLTE library by extending the baseline code to allow LTE transmissions in the unlicensed bands and coexistence with WiFi. We also expand previous results on this emerging research area by showing how different vendor-specific mechanisms in WiFi cards might affect coexistence.

I. INTRODUCTION

The paper motivates an open-source LTE SDR platform because existing testbeds are difficult to customize, instrument, or extend. srsLTE and srsUE address this need and support real-hardware experimentation with unlicensed LTE/WiFi coexistence.

  • Motivation: LTE testbeds commonly use expensive commercial hardware whose functionality and internals are difficult to modify or instrument.Research may require standard extensions, simplified protocols, or measurements across the PHY-to-IP stack.
  • Motivation: Open-source SDR testbeds trade some performance or capability for flexibility, openness, and easier stack instrumentation.Access to application code allows researchers to extract required metrics directly from the implementation.
  • Contribution: The paper presents srsLTE, an open-source LTE library, and srsUE, a complete software-radio LTE UE.The work describes their architecture, evaluates computational efficiency, and assesses their suitability for LTE experimentation.
  • Contribution: The platform is applied to duty cycle-based unlicensed LTE access and coexistence experiments with WiFi.The study examines LTE silent periods and vendor-specific WiFi-card behavior that may affect WiFi performance.
  • Contribution: Real testbed experimentation is motivated by the need to evaluate emerging coexistence mechanisms under deployment-relevant conditions.The article extends prior PHY-only results by considering LTE silent periods and WiFi-card vendor differences.

II. SRSLTE: AN OPEN-SOURCE LTE LIBRARY FOR SDR

srsLTE is a modular, open-source PHY library for LTE Release 8 that emphasizes code reuse, portability, and extensibility. Its hierarchical modules and supported features provide building blocks for LTE applications and experiments.

  • A. Description: srsLTE is an open-source ANSI C library for the LTE Release 8 PHY layer, designed for modularity, code reuse, and minimal dependencies.It operates on sample buffers in system memory and supports Ettus USRP devices through UHD interfaces.
  • Features: The library supports LTE Release 8 FDD operation, single-antenna and transmit-diversity modes, and UE cell search and synchronization.These capabilities are listed among the current library features.
  • Features: srsLTE provides downlink and uplink LTE channel and signal support for UE and eNodeB functions.Supported examples include PSS, SSS, PBCH, PDCCH, PDSCH, PRACH, PUSCH, PUCCH, and SRS.
  • Features: The library includes optimized turbo decoding, SIMD-related performance support, and MATLAB/OCTAVE MEX generation.The turbo decoder is reported at +100 Mbps with Intel SSE4.1/AVX and +25 Mbps in standard C.
  • Architecture: Researchers can customize, improve, or replace individual components without affecting the rest of the code.Modules are organized hierarchically into core processing, physical channels, UE processes, and example applications.

B. Computational Efficiency

srsLTE targets real-time SDR processing on general-purpose processors through lookup tables, precomputation, and data or instruction parallelism. Measurements indicate that one processing core can handle the complete PDSCH chain in real time.

  • B. Computational Efficiency: Computational efficiency is challenging in SDR because LTE receivers are substantially more complex than transmitters.srsLTE initially targets general-purpose processors and uses memory-intensive optimizations such as lookup tables and pre-generated signals.
  • B. Computational Efficiency: Lookup tables and precomputation reduce repeated computation for scrambling sequences, reference signals, PUCCH signals, CRC, encoding, and interleaving.These optimizations exploit inexpensive memory on general-purpose processors.
  • B. Computational Efficiency: SIMD and other parallel-processing techniques target the receiver’s most expensive modules, including turbo decoding, channel estimation, equalization, demodulation, and Viterbi decoding.VOLK selects architecture-specific kernels at runtime, while compiler intrinsics provide additional generic and SIMD versions.
  • B. Computational Efficiency: 1 processing core can process the entire PDSCH chain in real time for the tested LTE 1 ms data-arrival interval.The measurement used an Intel Core i7-3540M 3 GHz CPU and included OFDM demodulation, channel estimation, detection, rate recovery, turbo decoding, and CRC checking.

III. SRSUE: A COMPLETE UE SDR IMPLEMENTATION

srsUE is a software-radio LTE UE spanning PHY to IP and adding instrumentation, configuration, and networking capabilities. Its threaded architecture pipelines subframe processing to meet LTE latency requirements.

  • III. SRSUE: A COMPLETE UE SDR IMPLEMENTATION: srsUE covers all LTE network-stack layers from PHY to IP and achieves more than 60 Mbps downlink in a 20 MHz SISO configuration.The throughput was measured on an Intel Core i7-4790 against an Amarisoft LTE 100 eNodeB.
  • Features: srsUE supports configurable operation, network attachment, virtual network interfaces, authentication, detailed logs, packet captures, and trace metrics.These features support both user configuration and inspection of protocol behavior.
  • Research Use: Instrumentation and Wireshark capture expose LTE-stack behavior for cross-layer analysis, education, and prototype development.Traces can connect user experience, such as video streaming or web surfing, with signal quality and internal stack metrics.
  • Architecture: The UE organizes stack layers into classes with clean interfaces for message passing between layers.Separate threads handle PHY synchronization and DSP, MAC procedures and timers, and transport-buffer readers.
  • Architecture: LTE’s 4 ms response requirement leaves the UE 3 ms to decode PDSCH, generate ACK/NACK-related PUSCH, and send samples to the converter.The architecture therefore pipelines processing associated with each subframe rather than dividing work only into uplink and downlink threads.

COEXISTENCE SCENARIO

The srsLTE platform is used to evaluate coexistence between unlicensed LTE and WiFi on real hardware, capturing indoor propagation, capture, and carrier-sense effects that simulations may model inadequately.

  • COEXISTENCE SCENARIO: The evaluation studies LTE and WiFi links operating in the same unlicensed band.Real hardware enables assessment of complex indoor propagation effects and issues such as capture and carrier sense.

A. LTE/WiFi Testbed

The LTE/WiFi testbed combines an Ettus USRP-based LTE station with WiFi nodes and two vendor-specific wireless NICs to expose hardware-dependent coexistence behavior.

  • A. LTE/WiFi Testbed: The LTE station uses an Ettus USRP B210 connected over USB 3.0 to a Linux PC running srsLTE.The LTE base station is 34 cm from the WiFi transmitter and 35 cm from the receiver, while the WiFi link is 94 cm long.
  • A. LTE/WiFi Testbed: The experiments repeat measurements with Atheros AR9390 and Broadcom BCM4321 WiFi NICs to detect vendor-specific performance issues.The two cards are used to identify differences attributable to 802.11 hardware mechanisms.

B. Impact of duty cycle

The experiments vary LTE activity using a randomized duty cycle and show that WiFi throughput generally decreases with LTE activity, with robustness depending on MCS, band, and interference level.

  • B. Impact of duty cycle: The duty-cycle experiment measures normalized WiFi throughput over repeated 10 s trials using box-and-whisker summaries.The normalized value is the measured throughput divided by the maximum throughput at 0% LTE duty cycle.
  • B. Impact of duty cycle: WiFi throughput is inversely proportional to LTE duty cycle when LTE active and silent periods are relatively long.The silent period can accommodate several WiFi packets.
  • B. Impact of duty cycle: Lower WiFi MCS increases transmission robustness, while the effect is minor when LTE interference is negligible or overwhelming.In the 2.4 GHz band, the 54 Mbps MCS is especially sensitive to LTE interference.
  • B. Impact of duty cycle: Randomized LTE scheduling with relatively long on and off periods produces the expected qualitative impact for both WiFi cards.The study then examines MCS, transmission power, and frequency band as additional performance factors.

C. Impact of transmission power

With LTE duty cycle fixed at 50%, the study varies transmitter power and finds that WiFi power has little effect, while LTE power can strongly affect throughput and resource sharing.

  • C. Impact of transmission power: WiFi transmission power has little impact on performance in the small testbed.This indicates good WiFi-link signal quality across the tested configurations.
  • C. Impact of transmission power: Lower WiFi MCS increases normalized throughput, suggesting that many WiFi transmissions collide with LTE rather than merely deferring.More robust transmission schemes are therefore favored under the observed interference.
  • C. Impact of transmission power: The Atheros card shows a small but consistent improvement from 8 dBm to 17 dBm when MCS is 54 Mbps and LTE power is -6 dBm.This observation further supports the role of collisions in the measured performance variation.
  • C. Impact of transmission power: Transmission-power tuning could adjust resource sharing between WiFi and LTE, but the steep LTE-power response and minor WiFi-power effect may limit practicality.The authors identify power tuning as potentially useful but not necessarily practical.

D. Impact of bandwidth used

Varying LTE bandwidth can alter WiFi throughput differently across wireless cards, indicating that bandwidth-based coexistence control may be vendor-dependent.

  • At -16 dBm, WiFi throughput varies considerably with LTE PRB count, whereas at 12 dBm the number of PRBs does not affect throughput.
  • Atheros and Broadcom cards exhibit opposite throughput patterns as LTE bandwidth increases, depending on MCS and transmission power.Broadcom throughput decreases with greater interference for 54 Mbps MCS, while Atheros throughput increases with PRB count in the reported case.
  • Different proprietary clear channel assessment mechanisms are proposed as the explanation for the WiFi cards’ contrasting responses.
  • Changing LTE PRB count may help manage LTE/WiFi coexistence but can produce vendor-dependent performance and potential fairness issues in multi-vendor environments.

E. Impact of the central frequency

LTE central-frequency placement affects WiFi throughput according to spectrum overlap, transmit power, MCS, and WiFi hardware. The results support spectrum tuning while exposing hardware-dependent coexistence behavior.

  • At 12 dBm, large LTE/WiFi spectrum overlap causes substantial interference for both cards, while ±20 MHz offsets reduce it.The interference pattern is practically symmetrical around the WiFi center frequency.
  • At -16 dBm, Broadcom throughput generally falls as frequency offset narrows for 54 Mbps MCS, while robust MCS performance changes little.Atheros throughput is maximised either at a 20 MHz offset or when LTE and WiFi use the same center frequency.
  • At -1 dBm, an offset of 15 MHz causes a remarkable throughput drop for Broadcom cards using 6 Mbps MCS.
  • The measurements confirm that LTE/WiFi coexistence depends on WiFi hardware and that spectrum tuning may improve coexistence.
  • The srsLTE platform enables further experimental validation of coexistence mechanisms whose fairness effects may be difficult to predict from analysis or simulations.
Loading 1602.04629v1…