Vision and Goals
The WiPLASH project aims to take a radical step further in computing by designing a new breed of massive heterogeneous architectures with accelerator-like performance, but without loss of purpose generality. Towards this end, WiPLASH will pioneer an on-chip wireless communication based on graphene terahertz nano-antennas able to provide architectural plasticity, reconfigurability and adaptation to the application requirements. See our position paper for more details.
To this end, the project aims to achieve three specific objectives:- Prototype a miniaturized and tunable graphene antenna in the terahertz band
- Co-integrate graphene RF components with submillimeter-wave transceivers
- Demonstrate low-power reconfigurable wireless chip-scale networks
The culminating goal is to demonstrate that the wireless plane offers the plasticity required by future computing platforms by speeding up at least one key application by 10X over a state-of-the-art baseline. This will demonstrate that our approach can pave the way for a new generation of scalable, massively parallel biologically plausible AI processors.
Project Structure
The WiPLASH research project takes a vertical approach and touches upon different aspects of design from the implementation and integration of graphene antennas to the development of heterogeneous architectures based on wireless communications within package. Research is divided into seven work packages, five of which are technical work packages. We detail them below:
- WP1: RF Design and Implementation
WP1 develops the RF, mm-wave and THz device components necessary to experimentally assess the intra-chip and inter-chip wireless communication channels within a computing package. Design and fabrication of components within the 60-300 GHz range are planned, and support for the test of graphene-enabled heterogeneous and tunable solutions may be provided.
[Project vision] [Review on graphene-based RF circuits]
- WP2: Technological Integration
WP2 will be devoted to performing graphene integration on wafer scale with high performance required for the targeted application. Through this process, we aim to demonstrate that the graphene antenna and other graphene-based components (including diodes or larger RF components) can be co-integrated with silicon components forming the transceiver.
[Improving stability of tuning] [Terahertz rectennas] - WP3: Wireless Communications within Package
WP3 develops mechanisms for mmWave-THz wireless communications within package, tightly coupled to the capabilities of graphene antennas and to the requirements of the architecture. More specifically, the aims are to characterize the mmWave-THz wireless channel within a computing package and to develop a protocol stack that leverages the tunability of graphene antennas and provides channel allocation, scheduling, and routing schemes that dynamically adapt to the architecture.
[Towards spatial multiplexing] [MAC protocols for intra-chip communications] - WP4: Architecture Design
WP4 develops a massively parallel heterogeneous architecture for artificial intelligence, exploiting the two key innovation technologies (i.e. THz wireless channels and in-memory computing accelerators) to target deep neural networks as well as novel brain-inspired computing paradigms.
[DNN Inference with in-memory computing] [...and wireless interconnects] - WP5: Multi-scale Simulation
WP5 develops a novel heterogeneous simulation framework leveraging the processors and accelerators from WP4 and wireless interface from WP1/2/3. The proposed framework will enable system-level thermal, application, architecture-aware simulations of a wide range of applications, including but not limited to machine learning.
[Multi-core CPU systems with in-memory computing] [...and wireless interconnects] - WP6: Dissemination and Exploitation
WP6 deals with both (i) disseminating the project vision and results to different audiences, including publication of journal papers, presentations to conferences, lectures at undergraduate and graduate levels, talks in high schools, presence in social media, and others; and (ii) exploiting the results of the project through open-sourcing, licensing, patenting, or raising funding beyond the project - WP7: Management
WP7 is the coordination and management work package.
Progress
Towards its overarching and specific goals, the scientific work packages of WiPLASH focused on the following key actions during the first three years of execution:
- Establish the system specifications and the measurement setup. The specifications were decided based on current and future transceiver capabilities, as well as communication requirements from the architecture.
- Prototype the antenna. Theoretical analysis and simulations were used to anticipate the working point of the antennas depending on the graphene characteristics and antenna size, paving the way for a working antenna. Multiple prototypes with different signal sources, biasing schemes, antenna shapes, and integration capabilities have been fabricated to show the capability of graphene to generate a tunable emission in the terahertz band.
- Integrate the antenna with transceivers. We have developed RF transceiver circuits (also containing graphene RF devices) compatible with wafer-scale graphene production and integration, showing that there is a path to antenna-transceiver co-integration.
- Model the wireless channel within the computing package. Simulations in the range of 60-240 GHz have been performed assuming multiple computing packages. This allows us to perform accurate link budget analysis and other communication-related tasks. With this, we also demonstrated that a flip-chip package can support multiple frequency and space channels, a possibility that could be enabled by graphene-based antenna arrays.
- Develop communication protocols for the chip-scale scenario. We have worked on solutions at the physical and link layers that are able to manage multiple wireless channels concurrently. Simulations have shown significant improvements over single-channel protocols.
- Study the impact of wireless links in heterogeneous multi-chip architectures. We have considered domain-specific and general-purpose architectures, in both cases integrating in-memory computing cores, for different wireless communication speeds. We have observed substantial speedups already for links of several tens of Gb/s, compatible with the protocols and antenna specifications assumed at lower layers. This study is enabled by the augmentation of the full-system simulator of WiPLASH with ways to simulate heterogeneous architectures with in-memory computing cores and wireless communications within and across chiplets.
Given its disruptive nature, the wireless plasticity vision delivers competitive advantages to new and high-potential actors (young researchers, deep-tech SMEs) in a field that, traditionally, exhibits large barriers to entry. In that direction, WiPLASH continues to reach out to the industrial, scientific and general audiences explaining our progress and results. To this end, we participated in conferences, industrial fairs, debate panels, summer schools. We have prepared yearly newsletters and have given invited talks to high school students and interviews in general media. We have partnered up with several EU projects on similar and complementary technologies in 5G/6G networks, electronics and AI, as well as on gender issues.
For a selection of results, please check out our selected publications.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 863337.
Project Coordination: Sergi Abadal (UPC)
www.upc.edu