180 likes | 312 Views
MobileMAN Hardware development for the MAC module: current state and outlook by Ralph Bernasconi, Ivan Defilippis, Silvia Giordano, Alessandro Puiatti SUPSI Manno Cambridge, March 23 2004. Contents Introduction Steps done Where we are Next steps Data structure description
E N D
MobileMANHardware development for the MAC module: current state and outlookbyRalph Bernasconi, Ivan Defilippis, Silvia Giordano, Alessandro Puiatti SUPSI Manno Cambridge, March 23 2004 MobileMAN meeting, Cambridge
Contents • Introduction • Steps done • Where we are • Next steps • Data structure description • Cross-layering questions MobileMAN meeting, Cambridge
IntroductionProject’s SUPSI-DIE goals (reminder) • Provide flexible development- and test-bed for enhanced WLAN protocol (CNR) • Bursty responsive MAC • Better contention scheme (backoff); avoid collisions even for dense and active networks • Ensure compatibility with 802.11x • Provide few samples to partners • Provide inputs for new standard MobileMAN meeting, Cambridge
Steps Done (1/5) • Analysis of the Market for the 802.11 implementations • Which kind of solutions were present • Which solution best suited our needs • Cost • Reliability • Flexibility • Access to the software code • Access to the firmware • Easy development environment • Secure and prompt technical support MobileMAN meeting, Cambridge
Steps Done (2/5) • Analysis of the implementation starting point in relation with the market analysis • Starting from scratch: • Implementation fo all the chain: MAC - BB - RF • Starting from a “middle point“: • Implementation of the MAC only • Starting from an already implemented solution: • Change software and firmware for our pourpose MobileMAN meeting, Cambridge
Steps Done (3/5) • Choosed solution • Starting from a “middle point“ • BB and RF already developed by Elektrobit AG: • DT20 Modem • Ourselv MAC implementation on Texass Instrument DSP: • TI C6713 board MobileMAN meeting, Cambridge
Steps Done (4/5) • Analysis of 802.11 standard in deep • Procedures and flowchart for the TX/RX farther implementation • Implementations on a single board • CRC on FPGA • TX/RX between MAC and BB • Channel Sensig machanism on MAC, (MAC – RF) • MAC TX mechanism on DSP • MAC RX mechanism on DSP • Standard Backoff mechanism • Regular 802.11 frame generation • Fragmented frame generation MobileMAN meeting, Cambridge
Steps Done (5/5) • Duplicate above steps in a second board • Stress tests in TX between boards connected by cable • Stress tests in RX between boards connected by cable • Stress tests in TX/RX between boards connected by cable • Stress tests in TX/RX between boards in wireless mode MobileMAN meeting, Cambridge
Where we are • TX/RX of regular 802.11 frames OK • Backoff mechanism implemented • TX/RX of fragmented frame OK • RTS/CTS/DATA/AK handshake not yet implemented DT20 - Modems Level-Switch interfaces JTAG interface TI C6713 boards MobileMAN meeting, Cambridge
Next Steps • RTS/CTS/DATA/AK handshake • Test of Backoff standard • Implementation on a third board • Communication tests with three cards • Implementation of new Backoff algorithm • Communication between board and PC through FireWire interface • Implementation of Data structure MobileMAN meeting, Cambridge
Data Structure description (1/4) • Memory slotted in parameterized slot length (default frame max length) • Descriptor mechanism for handling data • Different queues for managing data transfer: • From MAC to PHY and from PHY to MAC • From MAC to Host and from Host to MAC • Implementation of priority mechanisms for each queue • Control flags in descriptor in which: • DATA = Frame • DATA = parameters for/from cross-layering MobileMAN meeting, Cambridge
Host to MAC Queue Begin End MPQ Descriptor x Host to MAC Queue Begin Descriptor x+1 Descriptor x+n+1 End HMQ End HMQ MAC h. y+n Frame body y+n End MPQ FCS y+n Data Structure description (2/4) Memory Data Area Descriptor y+n+1 MAC header x MAC h. y+n+1 Frame body x Frame body y+n+1 FCS x FCS y+n+1 Frame body x+1 Descriptor x+n MAC to PHY Queue MAC (DSP) Frame body x+n Frame body x+n+1 Descriptor y+n-1 Descriptor y+n MAC h. y+n-1 Frame body y+n-1 FCS y+n-1 Frame MobileMAN meeting, Cambridge
MAC to Host Queue Begin Descriptor x Descriptor x+1 PHY to MAC Queue PHY to MAC Queue End MHQ End MHQ MAC h. y+2 Frame body y+2 End PMQ FCS y+2 Data Structure description (3/4) Memory Data Area Frame body x Frame body x+1 Descriptor x+n MAC (DSP) Begin Frame body x+n Descriptor x+n+1 Descriptor y MAC header y Frame body x+n+1 Frame body y Descriptor y+1 Begin FCS y MAC h. y+1 Descriptor y+2 Frame body y+1 FCS y+1 Frame MobileMAN meeting, Cambridge
Descriptor 4 Pn Descriptor 2 Pn-1 Descriptor k Pn Descriptor g P0 Descriptor 3 Pn Descriptor h Pn-1 Descriptor 1 Pn Descriptor 2 P0 Descriptor 3 P0 Descriptor 2 Pn End Q-Pn Descriptor 1 Pn-1 End Q-Pn-1 Descriptor P0 End Q End Q-P0 Data Structure description (4/4) Management of queues with different levels of priority MobileMAN meeting, Cambridge
Cross-Layering Questions • How we think to implement the cross-layering: • A shared memory area in the host? • A s.m.a made in two blocks: one in the host and one in the MAC? • Other? • Who is the manager of the cross-layering data flow? • In which way the layers access the cross-layer area? • Which parameters we have to put in the cross-layer area? MobileMAN meeting, Cambridge
Questions? • Thanks for your attention! MobileMAN meeting, Cambridge