Personal tools

Difference between revisions of "VLSI Implementation of a Low-Complexity Channel Shortener for 2G EC-GSM-IoT and Evolved EDGE"

From iis-projects

Jump to: navigation, search
 
(7 intermediate revisions by the same user not shown)
Line 1: Line 1:
[[File:channel_shortening.png|thumb]]
+
[[File:channelShortening.png|thumb|A channel shortening filter transforms a long channel impulse response into a short impulse response.]]
  
==Short Description==
+
==Introduction==
Today's wireless devices have to deal with multi-path propagation caused by reflecting objects along the receive paths. These reflections lead to a long channel impulse response, which makes the channel equalization complex. Therefore channel shortening filters are used to transform the long channel impulse response into a shorter one. At IIS during the last years, several channel shortening algorithms and implementations have been presented.
+
Today's wireless devices have to deal with multi-path propagation caused by reflecting objects along the receive paths. These reflections lead to a long channel impulse responses, which makes the channel equalization complex. Therefore channel shortening filters are used to transform the long channel impulse response into a shorter one as illustrated in the figure. At IIS during the last years, several channel shortening algorithms and implementations have been presented. While dual-diversity streams have been introduced with the Evolved EDGE 2G cellular system [1], the recent EC-GSM-IoT standard achieves up to 20 dB coverage extension by means of up to 64 blind repetitions [2]. The blind repetitions can be seen as multiple diversity streams.
  
Your task in this project is to design and implement a hardware architecture of a novel channel shortening algorithm we developed recently [1]. You will get a MATLAB implementation of the algorithm as a reference implementation, from which you can start building your architecture. After you sketched a block diagram, you will implement the architecture in HDL. Synthesis and Place&Route using CAD tool lead to an ASIC implementation.
+
==Project Description==
 +
Just recently a Mutual Information Lower Bound (MILB) detector was published [3]. It combines multiple diversity streams before equalization, thus requiring a single stream equalizer only. A simplified block diagram of the MILB detector is depicted in Figure 2. It has I/Q samples of N diversity streams as input and outputs Log-Likelihood Ratios (LLR) at its output. The detector consists of channel estimation, variance estimation, shortening filter, and a Max-Log-MAP (MLM) equalizer. The overall goal of this project is the first ASIC realization of a MILB detector with support for the entire 2G GSM based cellular standard family ranging from extended coverage EC-GSM-IoT to high throughput Evolved EDGE. However, the work load would be beyond the scope of a semester project. Therefore, a floating-point, fixed-point, and HDL model of an MLM equalizer will be provided by the supervisors. Furthermore, a Matlab framework will be provided, as well. The students will implement the channel estimation, variance estimation, channel shortener, and glue logic to complement the provided MLM equalizer towards a full MILB detector. Synthesis and place and route of the entire detector conclude the targeted ASIC implementation.
  
===Status: In Progress ===
+
===Status: Completed ===
: Students: Robert Balas (sem16h28) and Georg Rütishauser (sem16h7)
+
: Students: sem16h28, sem16h7
 
: Supervision: [[:User:Weberbe|Benjamin Weber]], [[:User:Mkorb|Matthias Korb]]
 
: Supervision: [[:User:Weberbe|Benjamin Weber]], [[:User:Mkorb|Matthias Korb]]
 
+
: [http://asic.ethz.ch/2016/Pride_of_Babylon.html Chip gallery page of Pride of Babylon]
===Character===
 
: 30% Theory/Matlab
 
: 70% HDL, Synthesis, Place&Route
 
 
 
===Prerequisites===
 
: VLSI I
 
: Matlab
 
  
 
===Professor===
 
===Professor===
[http://www.iis.ee.ethz.ch/portrait/staff/huang.en.html Qiuting Huang]
+
[http://www.iis.ee.ethz.ch/people/person-detail.html?persid=78758 Qiuting Huang]
  
 
==References==  
 
==References==  
 +
[1] 3GPP. Release 7. http://www.3gpp.org/release-7, 2007.
 +
 +
[2] 3GPP. Release 13. http://www.3gpp.org/release-13, 2016.
  
[1] Hu, Sha, et al. "A Low-complexity Channel Shortening Receiver with Diversity Support for Evolved 2G Devices." ''IEEE International Conference on Communications''. 2016.
+
[3] Sha Hu, Harald Kröll, Qiuting Huang, and Fredrik Rusek. A Low-complexity Channel Shortening Receiver with Diversity Support for Evolved 2G Devices. In ''IEEE International Conference on Communications'', 2016.
  
 
[[Category:Digital]]
 
[[Category:Digital]]
 
[[Category:Semester Thesis]]
 
[[Category:Semester Thesis]]
[[Category:In progress]]
+
[[Category:Completed]]
 
[[Category:ASIC]]
 
[[Category:ASIC]]
 
[[Category:Telecommunications]]
 
[[Category:Telecommunications]]
 
[[Category:Weberbe]]
 
[[Category:Weberbe]]
 
[[Category:Mkorb]]
 
[[Category:Mkorb]]

Latest revision as of 09:39, 6 November 2017

A channel shortening filter transforms a long channel impulse response into a short impulse response.

Introduction

Today's wireless devices have to deal with multi-path propagation caused by reflecting objects along the receive paths. These reflections lead to a long channel impulse responses, which makes the channel equalization complex. Therefore channel shortening filters are used to transform the long channel impulse response into a shorter one as illustrated in the figure. At IIS during the last years, several channel shortening algorithms and implementations have been presented. While dual-diversity streams have been introduced with the Evolved EDGE 2G cellular system [1], the recent EC-GSM-IoT standard achieves up to 20 dB coverage extension by means of up to 64 blind repetitions [2]. The blind repetitions can be seen as multiple diversity streams.

Project Description

Just recently a Mutual Information Lower Bound (MILB) detector was published [3]. It combines multiple diversity streams before equalization, thus requiring a single stream equalizer only. A simplified block diagram of the MILB detector is depicted in Figure 2. It has I/Q samples of N diversity streams as input and outputs Log-Likelihood Ratios (LLR) at its output. The detector consists of channel estimation, variance estimation, shortening filter, and a Max-Log-MAP (MLM) equalizer. The overall goal of this project is the first ASIC realization of a MILB detector with support for the entire 2G GSM based cellular standard family ranging from extended coverage EC-GSM-IoT to high throughput Evolved EDGE. However, the work load would be beyond the scope of a semester project. Therefore, a floating-point, fixed-point, and HDL model of an MLM equalizer will be provided by the supervisors. Furthermore, a Matlab framework will be provided, as well. The students will implement the channel estimation, variance estimation, channel shortener, and glue logic to complement the provided MLM equalizer towards a full MILB detector. Synthesis and place and route of the entire detector conclude the targeted ASIC implementation.

Status: Completed

Students: sem16h28, sem16h7
Supervision: Benjamin Weber, Matthias Korb
Chip gallery page of Pride of Babylon

Professor

Qiuting Huang

References

[1] 3GPP. Release 7. http://www.3gpp.org/release-7, 2007.

[2] 3GPP. Release 13. http://www.3gpp.org/release-13, 2016.

[3] Sha Hu, Harald Kröll, Qiuting Huang, and Fredrik Rusek. A Low-complexity Channel Shortening Receiver with Diversity Support for Evolved 2G Devices. In IEEE International Conference on Communications, 2016.