The WIT910 radio transceiver provides reliable wireless connectivity for either
point-to-point or multipoint applications. Frequency hopping spread spectrum technology
ensures maximum resistance to noise and multipath fading and robustness in the presence of
interfering signals, while operation in the 900MHz ISM band allows license-free use and
worldwide compliance. Standard communication rates between the WIT910 and the host
are supported between 1200pbs and 57.6bps. Non-standard rates are supported as well. An
on-board buffer and an error-correcting over-the-air protocol provide smooth data flow and
simplify the task of integration with existing applications.
- Multipath fading impervious
frequency hopping technology
with 54 frequency channels
(902 to 927 MHz).
- Supports point-to-point or
multipoint applications.
- Transparent ARQ protocol
w/512byte buffer ensures data
integrity.
- Digital addressing supports up to
64 networks, with 62 remotes per
network.
1.1. Why Spread Spectrum?
The radio transmission channel is very hostile, corrupted by noise, path loss and
interfering transmissions from other radios. Even in a pure interference-free
environment, radio performance faces serious degradation through a phenomenon
known as multipath fading. Multipath fading results when two or more reflected rays of
the transmitted signal arrive at the receiving antenna with opposing phase, thereby
partially or completely canceling the desired signal. This is a problem particularly
prevalent in indoor installations. In the frequency domain, a multipath fade can be
described as a frequency-selective notch that
reflections change due to motion of the radio or objects within its range. At any given
time, multipath fades will typically occupy 1% - 2% of the band. This means that from
a probabilistic viewpoint, a conventional radio system faces a 1% - 2% chance of signal
impairment at any given time due to multipath.
Spread spectrum reduces the vulnerability of a radio system to interference from both
jammers and multipath fading by distributing the transmitted signal over a la rger region
of the frequency band than would otherwise be necessary to send the information. This
allows the signal to be reconstructed even though part of it may be lost or corrupted in
transit.
Narrowband vs. spread spectrum in the presence of interference
1.2. Frequency Hopping vs. Direct Sequence
The two primary approaches to spread spectrum are direct sequence (DS) and
frequency hopping (FH), either of which can generally be adapted to a given
application. Direct sequence spread spectrum is produced by multiplying the
transmitted data stream by a much faster, noise-like repeating pattern. The ratio by
which this modulating pattern exceeds the bit rate of the baseband data is called the
processing gain, and is equal to the amount of rejection the system affords against
narrowband interference from multipath and jammers. Transmitting the data signal
as usual, but varying the carrier frequency rapidly according to a pseudo-random
pattern over a broad range of channels produces a frequency hopping spectrum
system.
One disadvantage of direct sequence systems is that due to spectrum constraints and
the design difficulties of broadband receivers, they generally employ only a minimal
amount of spreading (typically no more than the minimum required by the regulating
agencies). For this reason, the ability of DS systems to overcome fading and in-band
jammers is relatively weak. By contrast, FH systems are capable of probing the
entire band if necessary to find a channel free of interference. Essentially, this
means that a FH system will degrade gracefully as the channel gets noisier while a
DS system may exhibit uneven coverage or work well until a certain point and then
give out completely.
Because it offers greater immunity to interfering signals, FH is often the preferred
choice for co-located systems. Since direct sequence signals are very wide, they
tend to offer few non-overlapping channels, whereas multiple hoppers may
interleave with less interference. Frequency hopping does carry some disadvantage
in that as the transmitter cycles through the hopping pattern it is nearly certain to
visit a few blocked channels where no data can be sent. If these channels are the
same from trip to trip, they can be memorized and avoided; unfortunately, this is
generally not the case, as it may take several seconds to completely cover the hop
sequence during which time the multipath delay profile may have changed
substantially. To ensure seamless operation throughout these outages, a hopping
radio must be capable of buffering its data until a clear channel can be found. A
second consideration of frequency hopping systems is that they require an initial
acquisition period during which the receiver must lock on to the moving carrier of
the transmitter before any data can be sent, which typically takes several seconds. In
summary, frequency hopping systems generally feature greater coverage and channel
utilization than comparable direct sequence systems. Of course, other
implementation factors such as size, cost, power consumption and ease of
implementation must also be considered before a final radio design choice can be
made.
As discussed above, frequency hopping radios periodically change the frequency at
which they transmit. In order for the other radios in the network to receive the
transmission, they must be listening to the frequency over which the current transmission
is being sent. To do this, all the radios in the net must be synchronized and must be set to
the same hopping pattern.
In point-to-point or point-to-multipoint arrangements, one radio module is designated as
the base station. All other radios are designated remotes. One of the responsibilities of
the base station is to transmit a synchronization signal to the remotes to allow them to
synchronize with the base station. Since the remotes know the hopping pattern, once they
are synchronized with the base station, they know which frequency to hop to and when.
Every time the base station hops to a different frequency, it immediately transmits a
synchronizing signal.
When a remote is powered on, it rapidly scans the frequency band for the synchronizing
signal. Since the base station is transmitting over 54 frequencies and the remote is
scanning 54 frequencies, it can take several seconds for a remote to synch up with the
base station.
Once a remote has synchronized with the base station, it must request registration from
the base station. The registration process identifies to the base station the remotes from
which transmissions will be received and not discarded. Registration also allows tracking
of remotes entering and leaving the network. The base station builds a table of serial
numbers of registered remotes. To improve efficiency, the 24-bit remote serial number is
assigned a 6-bit “handle” number. Two of these are reserved for system use, thus each
base station can register 62 separate remotes. This handle is how user applications will
know the remotes. Note that if a remote leaves the coverage area and then re-enters, it
may be assigned a different handle.
To detect if a remote has gone offline or out of range, the registration must be “renewed”
once every 256 hops. Registration is completely automatic and requires no user
application intervention. When the remote is registered, it will receive several network
parameters from the base. This allows the base to automatically update these network
parameters in the remotes over the air. Once a parameter has been changed in the base, it
is automatically changed in the remotes. The parameters auto matically changed are hop duration and the duty cycle.
At the beginning of each hop, the base station transmits a synchronizing signal. After the
synchronizing signal has been sent, the base will transmit any data in its buffer unless
data transmit delay has been set. The data transmit delay parameter allows for the
transmission of groups of continuous data in transparent mode (protocol mode
amount of data that the base station can transmit per hop is determined by the base slot
size parameter. The maximum amount of data sent by a base station per hop is 208 bytes.
If there is no data to be sent, the base station will not transmit until the next frequency.
The operation for remotes is similar to the base station without the synchronizing signal.
The amount of data a remote can send on one hop is dependent upon the hop duration,
the base slot size and the number of registered remotes. 212 bytes per hop is the
maximum data length a remote can transmit per hop, subject to limitations imposed by
the hop duration, the base slot size and the number of registered remotes. A detailed
explanation of this relationship is provided in Section 2.2.3. Minimum data length and
data transmit delay operate the same as with the base station.
Except for the registration process which occurs only when a remote logs onto the
network, the whole procedure is repeated on every frequency hop. Refer to the section
on Modem Commands for complete details on parameters affecting the transmission of
data.
2.2. Data Transmission
The WIT910 supports two network configurations: point-to-point and point-tomultipoint. In a point-to-point network, one radio is set up as the base station and the
other radio is set up as a remote. In a point-to-multipoint network, a star topology is used
with the radio set up as a base station acting as the central communications point and all
other radios in the network set up as remotes. In this configuration, all communications
take place between the base station and any one of the remotes. Remotes cannot
communicate directly with each other. It should be noted that point-to-point mode is a
subset of point-to-multipoint mode and therefore there is no need to specify one mode or
the other.
2.2.1. Point-to-Point
In point-to-point mode, unless data transmit delay or minimum data length have been set,
the base station will transmit whatever data is in its buffer limited to 208 bytes or as
limited by the base slot size. If the base station has more data than can be sent on one
hop, the remaining data will be sent on subsequent hops. In addition to the data, the base
station adds some information to the transmission over the RF link. It adds the address of
the remote to which it is transmitting, even though in a point-to-point mode there is only
one remote. It also adds a sequence number to identify the transmission to the remote.
This is needed in the case of acknowledging successful transmissions and retransmitting
unsuccessful transmissions. Also added is a 24-bit CRC to allow the base to check the
received transmission for errors. When the remote receives the transmission, it will
acknowledge the transmission if it was received without errors. If no acknowledgment is
received, the base station will retransmit the same data on the next frequency hop.
In point-to-point mode, a remote will transmit whatever data is in its buffer up to the limit
of its maximum data length. If desired, minimum data length and
also be set, which force the remote to wait until a certain amount of data is available or
the specified delay is exceeded before transmitting. If the remote has more data than can
be sent on one hop, it will send as much data as possible as a packet, adding its own
address, a packet sequence number and 24-bit CRC. These additional bytes are
transparent to the user application if the protocol mode is
00H (which is the default). In
the event a remote has more data to send, the data will be sent on subsequent hops. If the
transmission is received by the base station without errors, the base station will
acknowledge the transmission. If the remote does not receive an acknowledgment, it will
retransmit the data on the next frequency hop. To the user application, acknowledgments
and retransmissions all take place behind the scenes without the need for user
intervention.
The WIT910 has a point-to-point direct mode which fixes the remote radio’s handle at
30H. This mode is recommended for point-to-point applications, especially if the remote
is likely to periodically leave and re-enter the coverage area of the base. See the section
on Network Commands for details of this mode.
2.2.2. Point-to-Multipoint
In point-to-multipoint mode, data sent from the user application to the base station must
be packetized by the user application unless the remote device can distinguish between
transmissions intended for it and transmissions intended for other remote devices. This is
necessary to identify the remote to which the base station should send data. When the
user packet is received by the remote, if the remote is in transparent mode (protocol mode
0), the packetization bytes are stripped by the remote. In this instance the remote host
receives just data. If the remote is not in transparent mode, the remote host will receive
the appropriate packet header as specified by the remote’s protocol mode. Refer to the
section Protocol Modes for details on the various packet formats.
When a remote sends data to a base station in point-to-multipoint mode, the remote host
does not need to perform any packetization of the data. Remotes can operate in
transparent mode even though the base is operating in a packet mode. The remote will
add address, sequence and CRC bytes as in the point-to-point mode. When the base
station receives the data, the base station will add packetization header bytes according to
its protocol mode setting.
2.2.3. Handle Assignment
Handles are used to reduce overhead by not sending the unique 24-bit serial number ID
of a remote when sending or receiving data. The use of the various protocol modes causes
the base radio to issue CONNECT packets when a new remote registers with the base. In
addition to indicating the presence of a new remote, the CONNECT packets provide the
current relationship between remote serial numbers and handles.
When a remote links to a base and requests registration, it requests by default that it be
assigned handle
30H. This default request can be changed by the Set Default Handle
command. If that handle is not currently in use by another remote, the base will assign
that handle to the remote. If the requested handle is already in use by another remote, the
base will assign the next higher handle that is available. Thus, if remote requests handle
30H and that handle is already assigned, the base will assign the remote handle 31H if that
is available. If
31H is already assigned, the base will assign handle 32H is that is available
and so on.
When a remote leaves the coverage area of the base or otherwise loses link, e.g. the
remote was turned off or put into sleep mode, the base detects this event when the remote
does not renew its registration within 255 hops. With the default setting of 25msec per
hop, this could be as along as 6.38 seconds. If within this time the remote re-establishes
link with the base, the previous handle assigned to this remote will still be marked active
in the base radio. Thus the remote will be assigned a new handle. If the base radio is in
one of the protocol modes, a new CONNECT packet will be issued indicating the current
handle assigned to the remote. The remote is identified by the serial number that is
contained in the CONNECT packet.
If the radio is to be used in a point-to-point mode where there is only one base and one
remote, using the point-to-point mode command of the radios will override this handle
mechanism and always assign the remote the same handle.
2.2.4. TDMA Operation
For applications needing guaranteed bandwidth availability, the TDMA operation of the
WIT910 can meet this requirement. In the WIT910 TDMA scheme, each remote has an
assigned time slot during which it can transmit. The base station time slot is set
independently of the remote time slots through the Set Base Slot Size command. The
base station assigns each remote a time slot and informs the remotes of the size of the
time slot. All remote time slots are the same size that is determined by the number of
remotes registered with the base station. The slot size is a dynamic variable that changes
as the number of registered remotes changes. The remotes are continually updated with
the time slot size. This approach continually maximizes the data throughput. The base
station divides the amount of time available per hop by the number of registered remotes
up to a maximum of 16 times slots per hop. If the number of registered remotes is greater
than 16, the time slots will be spread across the required number of hops. For networks
with more than 16 possible remotes, the Set Duty Cycle command must be used to specify
a duty cycle -- the number of hops over which the time slots must be spread. For 1 to 16
remotes, no duty cycle is required; for 17 to 32 remotes a duty cycle of at least ½ is
required; and for 33 to 62 remotes a duty cycle of ¼ or more is necessary. An added
benefit of using the power save mode to set a duty cycle is improved average current
consumption efficiency. Refer to the Status Commands section for details of this
command.
When setting up a network, keep in mind that time slot length, maximum packet size and
hop duration are all interrelated. The hop duration parameter will determine the time slot
size and the maximum amount of data that can be transmitted per hop by the remotes.
There is a hard limit of the absolute maximum amount of data that can be sent on any
given hop of 212 bytes regardless of any parameters. (Note that this is different than the
208 byte maximum for the base station.) The base station requires 7.04 ms overhead for
tuning, the synchronization signal and parameter updating, as well as 1.11 ms overhead
for each remote. Thus the amount of time allocated per remote slot is roughly:
hop duration – base slot – 7.04ms - ( # of registered remotes)·1.11ms
( # of registered remotes)
Take for example a network comprised of a base station and 5 remotes. A hop duration
of 25 ms is chosen. We decide that the base station needs to be able to send up to 32
bytes each hop (equivalent to a capacity for the base of 19.2 kbps asynchronous).
Counting the 7.04 ms overhead for the base packet and making use of the fact that our RF
rate is 172.8 kbps, we determine that the base slot requires approximately:
32·8
172.8kbps
+ 7.04 ms = 8.52 ms
Each remote time slot will be:
25 ms – 8.52 ms – (5)·1.11 ms
5
= 2.18 ms
From our RF data rate of 172.8kbps we see that it takes 46.3 µs to send a byte of data, so
each remote will be able to send up to
= 47 bytes of data per hop.
2.18 ms
4
However, the WIT910 sends data in groups of 4 bytes. Thus, each remote will be able to
send 44 bytes of data. Note that the 44 bytes is the actual number of data bytes that can be
sent. If the WIT910 is using a protocol mode, the packet overhead does not need to be
considered. So in this example, the total capacity per remote would be:
44 bytes
25 ms
= 14.08 kbps
It is also useful to remember that the asynchronous data input to the WIT910 is stripped
of its start and stop bits during transmission by the radio, yielding a "bonus" of 10/8 or
25% in additional capacity. Thus, 1.25 x 14.08 kbps = 17.6 kbps asynchronous. In actual
deployments, some allowance must be made for retransmissions of data, yielding a
throughput somewhat less than the calculated value.
The above calculations are provided as a means of estimating the capacity of a multipoint
WIT910 network. To determine the precise amount of capacity, you can actually set up
the radio system and then query the maximum data length from one of the remotes in
control mode to discover its exact setting. Divide this number by the hop duration as
above to get the remote's exact capacity.
2.2.5. Full Duplex Communication
From an application perspective, the WIT910 communicates in full duplex. That is, both
the user application and the remote terminal can be transmitting data without waiting for
the other to finish. At the radio level, the base station and remotes do not actually
transmit at the same time. If they did, the transmissions would collide. As discussed
earlier, the base station transmits a synchronization signal at the beginning of each hop
followed by a packet of data. After the base station transmission, the remotes will
transmit. Each base station and remote transmission may be just part of a complete
transmission from the user application or the remote terminal. Thus, from an application
perspective, the radios are communicating in full duplex mode since the base station will
receive data from a remote before completing a transmission to the remote.
2.2.6. Error-free Packet Transmission Using ARQ
The radio medium is a hostile environment for data transmission. In a typical office or
factory environment, 1% - 2% of the 900MHz frequency band may be unusable at any
given time at any given station due to noise, interference or multipath fading. For
narrowband radio systems (and also many spread spectrum radio systems which use
direct sequence spreading), this would imply a loss of contact on average of over 30
seconds per hour per station. The WIT910 overcomes this problem by hopping rapidly
throughout the band in a pseudo-random pattern. If a message fails to get through on a
particular channel, the WIT910 simply tries again on the next channel. Even if two thirds
of the band are unusable, the WIT910 can still communicate reliably.
Data input to the WIT910 is broken up by the radio into packets. A 24-bit checksum is
attached to each packet to verify that it was correctly received. If the packet is received
correctly, the receiving station sends an acknowledgment, or
station. If the transmitter doesn't receive an
ACK, at the next frequency hop it will attempt
ACK, back to the transmitting
to send the packet again. When ARQ is enabled, the transmitting radio will attempt to
send a packet packet attempts limit times before discarding the packet. A value of
00H
disables ARQ. When it is disabled, any transmission received with errors is discarded. It
is the responsibility of the user application to track missing packets. A second parameter,
ARQ Mode, allows the choice between using ARQ to resend unsuccessful transmissions
or always sending a transmission packet attempts limit times regardless of the success or
failure of any given transmission.
All of this error detection and correction is transparent to the user application. All the
user application sees is error-free data from the modem. However, if the ARQ mode is
disabled, transmissions with errors are discarded, and missing data detection will be the
responsibility of the user application. Refer to the Protocol Commands section for
complete details.
2.3. Modes of Operation
2.3.1. Control and Data Modes
The WIT910 has two modes of operation: Control mode and Data mode. When in
Control Mode, the various radio and modem parameters can be modified. When in Data
Mode, only data can be transmitted. The default mode is Data Mode. There are two
ways to enter Control Mode. The first way is to assert the Configure (CFG) pin on the
modem. Upon entering Control Mode, the modem will respond with a > prompt. After
each command is entered, the modem will again respond with a > prompt. As long as the
CFG pin is asserted, data sent to the modem will be interpreted as command data. Once
the CFG pin is de-asserted, the modem will return to Data Mode.
The second method for entering Control Mode is to send the escape sequence
(all lower case) followed by a carriage return. In the default mode, the escape sequence is
only valid immediately after power up or after de-assertion of the Sleep pin on the
modem. The modem will respond in the same way with a > prompt. To return to Data
Mode, enter the Exit Modem Control Mode command,
Sleep pin. There are three modes for the escape sequence, controlled by the Set Escape Sequence Mode command,
zc = 0 Escape sequence disabled
zc = 1 Escape sequence available once at startup (default setting)
zc = 2 Escape sequence available at any time
zc2 mode setting is useful if the user application has a need to change the modem
The
settings "on the fly". In this mode the escape sequence is always enabled and may be sent
at any time after a pause of at least 20ms. The modem will respond in the same way as
when in the default mode. It is necessary to issue the Exit Modem Control Mode
command,
z>, before resuming data transmission.
Note: The escape sequence must be interpreted as data until the last character is received
and as such may be transmitted by the modem to any listening modems.
2.3.2. Sleep Mode
To save power consumption for intermittent transmit applications, the WIT910 supports a
Sleep Mode. Sleep Mode is entered by asserting the Sleep pin on the modem interface.
While in Sleep Mode, the modem consumes less than 250 µA. This mode allows the
radio to be powered off while the terminal device remains powered. After leaving Sleep
Mode, the radio must re-synchronize with the base station and re-register.