4GWE: Development Tools

The mathematical theory often associated with wireless and mobile physical layer algroithms are complex and require a lot of thought when looking for new solutions in the wireless network. OFDM and MIMO are at the base of these algorithms. These sessions provides an intuitive and straightforward view beneficial to a wide audience of engineers and project managers.


After an overview of the discussion about how to design a 4G device from Frank Schirrmeister of Synopsys, Frank Vincze of Steepest Ascent talked about the library software.

The Library from Steepest Ascent deals with issues of channel optimisation, MIMO, and other LTE device management charteristics.

Operations that are needed to manage the development of the device in a set of libraries that take you through the OSI stack. 

Frequency Division Duplexing and the transciever functions are all within the library. 

The library has prebuilt models that represent some standard models of design.  This allows product differentiation to be done in relation to some standard sets that include test scenarios.  All of this done within the Math Kernel Library within System Studio.

Frequency is the subcarrier which makes up the orthogonal portion of the  OFDM strategy.

LTE uses a number of channel coding techniques in LTE.

The channel that makes the bitstream is controlled with a blended data channel and the control channel. 

All this signal processing equates to the throughput of the device. 

The algorithm in the MLK are put into the parameter blocks represented in software to enable the developer to manage the design through a GUI tool.

In LTE there are a number of reserved spaces in the packet format of the signal.

The blocks are the Scrambler, Symbol Modulation,  Layer Mapper and Precoder.  After your precode is completed it gets added to the LTE reserved signalling formats.

By using the libraries you get the benefit of testing against real data. So your can develop in short time frames.

The point of the Libraries "Don't Sweat the Math"

TenSilica not surprisingly is looking at how to move the design to silica.

After taking the libraries and developing your strategies you are to think about what goes on silica.

There is always a trade off between the energy savings of moving to silica and the flexibility of software.  This is the complex integration that a developer faces in designing the device.

Performance can be achieved in a number of ways with both hardware and software.

TenSilica has built a LTE Reference Architecture including some new techniques in turbo processing.

The speed of LTE's data path (150 MB/S) requires intense computation.

The issues that a designer faces in moving to LTE.
- is looking at existing legacy 3G design. Do they intend to be computational in design.
- to what extent does your exist model stand up to the new efficiency requirements.
- the DSPs embedded in their design may no longer be valid

The implications of the business decisions is the carrier is looking for dual design systems, with and expectation that the efficiency of LTE will compel a single design in the future.

Tensilica's focus is on the data plane processors hardware cores with software that supports the solutions to enhance performance.  This represents a blend of DSP and CPU functions to support the performance.

TenSilica has a compatiable foundation that controls the system.  It in effect provides backward compatiablity.  The silica has further system functions embedded in the format. 

The user defines the rest of the design using TenSilica Instruction Extension [TIE] Language. 

TenSilica offers a wide range of Silica models from single MAC to 16 MAC design.

Tensilica enables processing with four domains, Signal, matrix, softbits, and control.

Presented by:   Frank Schirrmeister Director, Product Marketing, System-Level Solutions Synopsys, Inc   Chris Rowen Chief Technology Officer Tensilica   Frank Vincze Region Manager, Americas Steepest Ascent