Commit graph

687 commits

Author SHA1 Message Date
nvt-se 96eb20d235 forward arguments even if REXMIT is not set 2010-03-04 14:17:43 +00:00
dak664 c65e09ceb2 Remove compiler warnings 2010-03-02 15:41:23 +00:00
nifi 5e43a6998b Replaced old framer (framer_get/framer_set) with new NETSTACK style configuration. Added nullrdc-framer that is a nullrdc with framer support. 2010-03-01 13:30:21 +00:00
nifi 0c084b8c68 added random init of sequence number 2010-03-01 13:17:01 +00:00
adamdunkels 524625dfd2 Removed debug string names of packet attribtues 2010-02-28 20:20:19 +00:00
adamdunkels 410d860a93 Improved handling of neighbor phase estimation: instead of busy-waiting while waiting for neighbor's phase, the new code schedules a timer just before neighbor wakes up 2010-02-28 20:19:47 +00:00
adamdunkels e6fceb7200 Added MAC-layer return value to indicate that a packet transmission was deferred 2010-02-28 20:18:30 +00:00
adamdunkels ec928a2a19 Make max queueing time depend on retransmission timeout 2010-02-28 14:15:58 +00:00
adamdunkels e06acc4525 Prod the watchdog if waiting for along time 2010-02-28 14:15:16 +00:00
adamdunkels 74da90a06b Don't count collisions as transmissions 2010-02-28 10:07:17 +00:00
adamdunkels f95a6cce85 Added function for obtaining packet queue length 2010-02-28 09:20:32 +00:00
adamdunkels 8fe15f7b45 New packet attribute for specifiying MAC layer retransmissions; reorder and commented packet attributes 2010-02-28 09:19:43 +00:00
adamdunkels 970a9eb9c7 Increased neighbor lifetime to allow for longer neighbor announcement intervals 2010-02-28 09:19:04 +00:00
adamdunkels 7165a3866f Significant rework of the Contiki data collection protocol:
* the new version makes use of MAC-layer feedback so that bad paths
  can be identified quicker and then avoided.

* the new code uses transport layer ACKs that contain feedback from
  the collect protocol: when a packet cannot be forwarded due to lack
  of resources, the ACK contains a flag that indicates that the packet
  could not be forwarded. ACKs also contain the routing metric of the
  sender, which improves agility in face of rapid path changes.

* loop detection and management has been improved: with higher path
  metric agility, the system is more prone to short-lived routing
  loops. Instead of dropping looping packets, the new version adjusts
  the routing metric for the routes that exhibit loops so that the
  risk for future loops is reduced.

* make use of packet attributes to inform the MAC layer of how many
  times packets should be retransmitted.
2010-02-28 09:18:01 +00:00
adamdunkels b6b4941871 Made the number of MAC-layer retransmissions configurable on a per-packet basis, via a packet attribute 2010-02-28 08:35:16 +00:00
adamdunkels ed88f487c8 Improved detection of link-layer ACKs; increased the time ContikiMAC waits when it has detected a link-layer ACK; increased the number of silence periods required to detect a CCA false positive 2010-02-28 08:33:21 +00:00
adamdunkels fa1a96636e Bugfix: the compile-time check for UIP_LLADDR_LEN previously generated a run-time error, when it should have generated a compile-time error 2010-02-28 08:30:39 +00:00
adamdunkels ba0e50a7b7 Set the default number of MAC layer retransmissions for IPv6 packets to 2 2010-02-28 08:29:42 +00:00
joxe 0ab2bc9a31 upgraded sicslowpan to hc-06 - no 48 bit multicast compression support yet 2010-02-25 15:50:58 +00:00
nifi b50d47fb80 added missing function in sicslowmac_driver 2010-02-23 22:45:11 +00:00
nifi 1cfcc8a73a Removed the combined name of CSMA and NETSTACK_RDC. The name of NETSTACK_RDC can be accessed separately if needed. 2010-02-23 20:42:45 +00:00
nifi c20047f657 Changed the network stack to use different structures for the different layers (network_driver, mac_driver, rdc_driver, radio_driver). This helps to avoid incorrect configuration. 2010-02-23 20:09:11 +00:00
adamdunkels bb62749de7 Made debug printfs optional 2010-02-23 18:49:58 +00:00
adamdunkels 0ab2ba433f Bugfix: the wrong callback pointer was passed to higher layers. 2010-02-23 18:49:45 +00:00
adamdunkels ed52586e18 Rework of parts of the ContikiMAC code based on input from experimental results. When channel check is triggered because of a false positive, the new version checks the radio several times to make sure that there really is an incoming packet. Otherwise, the radio is switched off. Cleaned up and removed old unused code. Added a few temporary debugging mechanisms and functions. 2010-02-23 18:49:05 +00:00
adamdunkels ca0417a923 Added the MAC layer callback functionality so that the Rime network layer gets information from the MAC about the status of sent packets: were they successfully transmitted, or was there a collisions? How many retries were there? 2010-02-23 18:38:05 +00:00
adamdunkels f10eedab3e Added debug printout 2010-02-23 18:37:15 +00:00
adamdunkels 2d13fecea7 Improved backoff management: timer is now doubled on every transmission, until the using module calls the neighbor_discovery_start() function again. Don't cancel the transmission even if a lower value arrives (this fixes a problem with multi-hop ETX propagaion). 2010-02-23 18:36:50 +00:00
adamdunkels 0eade2886c Better loop handling: when a loop is detected, the neighbor to which the packet was originally sent gets an inflated ETX; made sure that collect without announcments work; increased timeouts, to work nicer in larger networks 2010-02-23 18:35:23 +00:00
adamdunkels a4d9d6651b Added the option to specify the old value to observers so that they can determine if the announcement should be sent out quicklier than otherwise 2010-02-23 18:32:44 +00:00
adamdunkels b746b7fc06 Reworked the Rime/Chameleon interface so that packets now always hit Rime before Chameleon. Chameleon is now only called by Rime. This makes packet sniffing easier, and allows for sniffing packets after their header has been processed by Chameleon, thereby making per-packet power profiling much easier to implement 2010-02-23 18:29:53 +00:00
adamdunkels d8cce42d28 Set the default radio driver to nullradio 2010-02-23 18:28:04 +00:00
adamdunkels 0b2b7bcdba Removed inadvertedly left-over include file 2010-02-18 23:52:34 +00:00
adamdunkels 5cf3b83733 Common initialization code for the netstack 2010-02-18 23:15:26 +00:00
adamdunkels f49ebfe1c4 Temporarily disabled rime route udp because it breaks with newest MAC interfaces 2010-02-18 21:53:08 +00:00
adamdunkels e34eb54960 A work-in-progress rework of the Contiki MAC and radio layers. The
main ideas are:

* Separates the Contiki low-layer network stack into four layers:
  network (e.g. sicslowpan / rime), Medium Access Control MAC
  (e.g. CSMA), Radio Duty Cycling RDC (e.g. ContikiMAC, X-MAC), and
  radio (e.g. cc2420).
* Introduces a new way to configure the network stack. Four #defines
  that specify what mechanism/protocol/driver to use at the four
  layers: NETSTACK_CONF_NETWORK, NETSTACK_CONF_MAC, NETSTACK_CONF_RDC,
  NETSTACK_CONF_RADIO.
* Adds a callback mechanism to inform the MAC and network layers about
  the fate of a transmitted packet: if the packet was not possible to
  transmit, the cause of the failure is reported, and if the packets
  was successfully transmitted, the number of tries before it was
  finally transmitted is reported.
* NULL-protocols at both the MAC and RDC layers: nullmac and nullrdc,
  which can be used when MAC and RDC functionality is not needed.
* Extends the radio API with three new functions that enable more
  efficient radio duty cycling protocols: channel check, pending
  packet, and receiving packet.
* New initialization mechanism, which takes advantage of the NETSTACK
  #defines.
2010-02-18 21:48:39 +00:00
adamdunkels f8c804e28b ntikiMAC: a new radio duty cycling mechanism that uses a combination
of link-layer and MAC-layer mechanisms to achieve a very low power
consumption: during idle listening the radio is switched off between
99.2% and 99.6% of the time. One unicast transmission typically adds
between 0.07% and 1.2% additional radio time, depending on packet
size.

ContikiMAC uses the standard IEEE 802.15.4 message format and adds no
additional headers.

ContikiMAC is simple: it uses periodic two-shot channel sampling
during idle listening to keep the radio on-time down. Transmissions
are done with repeated transmissions until a link-layer ACK is
received.
2010-02-18 21:26:15 +00:00
adamdunkels f5c48751b5 Reworked 802.15.4 header creation and parsing so that it works regardless of the size of the rimeaddr_t typedef 2010-02-18 21:00:28 +00:00
adamdunkels 2fbf21b2b3 Refactored the phase optimization functionality into a separate module 2010-02-18 20:58:59 +00:00
adamdunkels 515a391ee4 Removed left-over debug printfs 2010-02-15 23:31:05 +00:00
dak664 bb9afb4fb1 Remove compiler warning 2010-02-15 20:49:38 +00:00
adamdunkels 4bb0375251 Buffer name definition bugfix 2010-02-15 19:23:54 +00:00
adamdunkels 9bb149ca13 Bugfix by Raimondas Sasnauskas: TCP SYN duplicate receptions in the SYN_RCVD state was not correctly handled 2010-02-15 18:03:07 +00:00
adamdunkels d67d1f7340 Major bugfix, found by Raimondas Sasnauskas: the TCP periodic timer was not started when a new connection request arrived. If there were no other active TCP connections, this means that the SYNACK would never be retransmitted, potentially leading to a forever stale TCP connection. 2010-02-09 12:58:53 +00:00
adamdunkels 4a522c96dd Added debugging statements for packet drops 2010-02-08 21:59:49 +00:00
adamdunkels 2b6a651233 Added missing include, made debug statement more explicit 2010-02-08 21:59:21 +00:00
adamdunkels 2eba08e8f6 Made queuebuf statistics configurable, and off by default 2010-02-08 21:10:32 +00:00
adamdunkels b98f7d891f Bugfix: added missing queuebuf allocation 2010-02-06 20:50:24 +00:00
adamdunkels 671216c558 No need to clear the headers now that we explicitly save a queuebuf when sending fragments 2010-02-06 17:07:43 +00:00
adamdunkels b6647b5630 Turns out that with cxmac, clearing the header after sending a packet is not enough to make fragmentation work. Instead, we allocate a queuebuf and restore the packet after from the queuebuf after sending it. 2010-02-06 09:42:49 +00:00