WirelessIdeas

Hardware

Xbee

  • Pros:
    • old chip, we know roughly how to use it
  • Cons:
    • moderately slow
    • have to make the library ourselves
  • Stats:
    • UART at 115200 baud
    • wireless speed at 100 Kbps
    • 2.8V ~ 3.4V, 3V typical
    • Xbee:
      • TX ON: 40mA, RX ON: 50mA, TRX OFF: 0.2mA, sleep: 0.01microA
      • 0.0 dBm
      • 25mm x 28mm
      • ($19)
    • Xbee Pro:
      • TX ON: 215mA, RX ON: 55mA, TRX OFF: 0.2mA, sleep: 0.01microA
      • 18 dBm
      • 25mm x 33mm
      • ($32)

IPv6

  • Pros:
    • faster
    • premade TCP/UDP library contiki
      • can use standard networking functions
      • has built in layer 3 mesh network
    • with a routing bridge, it could communicate directly with anything on the internet (the packets are the same)
  • Cons:
    • multiple parts
    • a little larger than xbee
  • Stats:
    • SPI at 8 Mhz
    • wireless speed at 2 Mbps
    • 3V
    • 5mA avg
    • 30mm x 50mm
    • ($11.49)
  • Components:
    • Transceiver
      • Atmel AT86RF231
        • TI CC2520 as alternate (includes amp, doesn't do ipv6)
      • 1.8V ~ 3.6V, 3V typical
      • TRX ON: 12mA, TRX OFF: 0.4mA, sleep: 0.02microA
      • 5mm x 5mm
      • ($4.73)
    • Front End (amp)
      • TI CC2591
        • CC2590 is low power (10 dBm) option
      • 2V ~ 3.6V, 3V typical
      • TX ON: 112mA, RX ON: 4mA, TRX OFF: 0.1microA
      • 20 dBm
        • this means the range should be greater than the xbeePro
      • 4mm x 4mm
      • ($5.25)
    • Antenna
      • ANT-2.45-CHPCT-ND
      • +0.5dBi
      • 3W max
      • 1mm x 6.6mm
      • note: requires 30mm x 50mm silicon
        • 30mm x 30mm ground plane - can have other chips on top
        • 30mm x 20mm empty of all other traces, chips, etc
      • ($1.51)

802.11b/g (regular wifi)

  • These notes are here for when a bridge chip needs to be made
  • WiFly GXS chip solution
  • hacked router
    • WRT54G V2 running OpenWRT
      • has 2 serial ports that we'd have to connect headers to, then connect our zigbee to one of them
        • if we go with the new zigbee, we need to convert SPI to serial (maybe a small atmega?)
      • openwrt wrt54g page
  • usb wifi stick
    • connect directly to a robot to enable wifi when robot is on
    • could make a similar board without the avr and sensor package to work as a bridge

Software Protocol (Xbee) (--out of date--)

As an upgrade from the current token-ring/ad hoc mesh network used by ColoNet, a geodesic packet forwarding network utilizing the existing XBee modules is suggested. This network would allow robots to communicate over a large area where some robot nodes are too far apart for direct communication.

The current model assumes all robots will remain in range with their token neighbors. This is in conflict with the future goal of long range communication over a wide area. Care needs to be taken that the increased computational requirements for the network do not compromise the communication efficiency of the current model.

Methods of Accomplishment:

  • Broadcast at 50% or lower signal strength, adjusting as necessary
  • Measure incoming signal strength, choose strongest neighbor
  • Decentralized control to ensure no robot is left out of the network
  • Poll sensors at a low frequency to ensure optimal connectivity

Caveats:

  • Multi-directional structure incurs the risk of loops
  • Requires greater computation and processing time
  • Requires revision of existing packet protocol

Additional Thoughts:

In the current system, when a robot moves out of range or can not be reached it is eliminated from the network. Since eliminating a robot from the network sacrifices any data from that robot, a more robust design might require a robot to attempt to move back into range of other robots or maximize the power of its broadcast signal.