Image Processing Reference
In-Depth Information
example, if the receiver has received at least three erroneous copies of a packet, it can try to figure
out the original packet by applying bit-by-bit majority voting. There are other packet combining
techniques, for example equal-gain combining.
Sometimes the packet error probability (and therefore the need for retransmissions) can be
reduced by proper tuning of packet sizes. Intuitively, it is clear that larger packets are more likely
hit by errors than smaller ones. On the other hand, with smaller packets the fixed-size packet header
becomes more dominant and leads to increased overhead. If the transmitter has estimates of current
channel conditions it can choose the “appropriate” packet size giving the desired trade-off between
reliability and efficiency [].
24.6 Wireless Fieldbus Systems: State of the Art
Fieldbus systems are designed to deliver hard real-time services under harsh environmental con-
ditions. A wireless fieldbus [] should be designed to provide as stringent stochastic timing and
reliability guarantees as possible over wireless links. However, in most of the literature surveyed in this
section this issue is not addressed. Nonetheless, we discuss existing approaches for different popular
fieldbus systems.
24.6.1 CAN
As already described in Section .., the CAN system [] uses a priority arbitration protocol on
the MAC layer, which cannot be implemented directly on a wireless link. Some approaches have been
developed to circumvent this; here we discuss a centralized and two distributed solutions [].
The distributed WMAC protocol uses a CSMA/CA (carrier-sense-multiple-access with collision
avoidance) scheme with priority-dependent backoffs. A station wishing to transmit a packet uses a
carrier-sense mechanism to wait for the end of an ongoing packet transmission. Ater this, the station
picks a backoff time depending on the priority value of the current packet. he station listens on the
channel during the backoff time. If no other station starts transmission, the station assumes that
it has the highest priority and starts transmitting its own packet. Otherwise, the station defers and
starts over after the other packet has been finished. In another distributed scheme, the CAN message
priority value is mapped onto the channel using an on-off keying scheme []. A station transmits a
short burst if the current priority bit is a logical one, otherwise it switches into receive mode. If the
station receives any signal it gives up, otherwise it continues with the next bit. The priority bits are
considered from the most significant bit to the least significant bit. If the station is still contending
after the last bit, it transmits the actual data packet. his approach requires tight synchronization and
fast switching between transmit and receive modes of the radio transceiver, which is a problem for
certainWLANtechnologies.
The centralized RFMAC protocol leverages the fact that CAN belongs to the class of systems using
the real-time database communication model. Data items are identified by unique identifiers. Similar
to FIP/WorldFIP, all communications are controlled by a central station broadcasting the variable
identifiers and causing the producers of the corresponding data items to transmit the data.
24.6.2 FIP/WorldFIP
The FIP/WorldFIP fieldbus uses a polling table to implement a real-time database []. To couple
wired and wireless stations, in [] a wireless-to-wired gateway is introduced, serving as central
station for the wireless part. he wireless MAC protocol uses time division multiple access (TDMA),
and each TDMA slot is used to transmit one data item (also called process variable).
In the OLCHFA project, a prototype system integrating wired and wireless FIP stations has been
developed. This system worked in the . GHz ISM band using a DSSS physical layer []. The
