A-MACAnother MAC: is a slick implementation of a receiver-initiated MAC protcol exploiting address filtering and auto-acks provided by 802.15.4 radios. A-MAC is built on top of the backcast primitive in which a probe is followed by a set of (auto-)acks that do not interfer destructively, but allow the initiator to decode the super-imposed reponses as a valid ack causing it to stay awake to receive the subsequent data frame. In the case of multiple senders data frames may collide, which is resolved by specifying a larger contention window in the next probe. Due to the auto-acking, probing for an empty channel is extremely fast, yielding large energy savings. To increase throughput, the data transfer is done on a different channel than the probing.
AI-LMACAdaptive Information-centric and Lightweight MAC: Improves LMAC by allowing nodes to control more than one slot.The decisdion on how many slots each node should claim is left to the data gathering layer above.
B-MACBerkeley MAC: Improvement on Low-Power Listening (LPL) with application-level control over check time, back-off window size, and power-down policy (e.g, no sleeping at the sink). Advanced Clear Channel Assignment (CCA) to handle random noise.
BitMACBitMAC: layers the network as a spanning tree from the sink, and restricts communication to up/down traffic between children and their parent. Each parent serves as an access point computing the TDMA schedule from (1-bit) transmission requests made by its childeren at the start of the frame. Multiple parts of the tree can operate in parallel due to the use of multiple radio channels. The information propagation for slot and channel assignment is done efficiently (concurrently) using On-Off-Keyeing (OOK modulation) providing an OR-channel.
BMABit-Map-Assisted: A rotating duties scheme (based on LEACH) where nodes in turn act as an access point controlling the TDMA schedule. At the beginning of each frame nodes send 1 bit to the access point indicating wether or not they have data to transmit. Th elength of a frame depends on the number of active senders in the cluster.
BuzzBuzzBuzzBuzz: a small layer on top of the 802.15.4 physical layer that employs multiple headers per packet and FEC (Reed-Solomon) to handle external interference caused by WiFi traffic.
CrankshaftCrankshaft: is a TDMA-style protocol designed for dense networks. Crankshaft is receiver-based to limit overhearing, and slot selection is by node ID (modulo #slots/frame). Contention is resolved by senders-only through CSMA+Sift before the slot-owner(s) poll, much like SCP-MAC.
CSMA-MPSCSMA with Mimimum Preamble Sampling: Incremental improvement over WiseMAC. Uses the alternating Tx/Rx wakeup scheme from STEM to cut the average preamble length in halve.
CSMA/ARCRandomized CSMA with Adaptive Rate Control: A CSMA scheme designed to be energy efficient and provide fairness to routed-through traffic. Employs random waits and backoffs to handle contention. No additional signaling (not even an ACK).
DMACData gathering MAC: Addresses the latency overhead of S-MAC for the convergecast communication pattern, by staggering receive/send slots according to the level in the tree. Uses CSMA with ACKs to arbitrate between children, and schedules overflow slots whenever a msg is received (much like T-MAC's adaptive duty cycle).
f-MACframelet-MAC: A deterministic MAC layer that gives hard guarantees on message transfer delays and available bandwidth, by having nodes send each message as a sequence of N framelets. Careful selection of the framelet frequency of each node guarantees collison-free transmission of at least 1 framelet. f-MAC works best in static and sparsely populated networks.
FLAMAFLow-Aware Medium Access: Improves on TRAMA by removing the periodic traffic information exchange between two-hop neighbors. This information is now transmitted upon request only (e.g., when an application flow is established) making FLAMA more efficient.
LMACLightweight MAC: A self-organizing TDMA scheme in which each node owns a slot in a fixed-length frame, in which it may send data to any of its neighbors. As part of its header, a node broadcasts the occpied slots in its one-hop neighborhood, allowing new nodes to select a unique two-hop slot guaranteeing collison-free operation (hence, no ACKing). All nodes must listen in on all slots to check for incoming traffic and to maintain synchronization.
LPLLow Power Listening: Nodes periodically check with a single probe if the channel is clear, so they can power down immediately. If the channel is busy, however, they keep listening until a startsymbol is detected. This reduces idle listening overhead at the expense of sending out long preambles (which must be larger than the check iterval).
MMACMobility-adaptive MAC: a TDMA scheme in which traffic schedules are based on information about traffic loads as well as mobility patterns of individual nodes. This info is sent periodically to a cluster head, which relays all info to its nodes with a single broadcast. Each TDMA frame contains a scheduled access section (of variable length) and a random access section (the remainder).
PACTPower Aware Clustered TDMA: uses passive clustering to creatre a backbone communication network of clusterheads and gateway nodes. These (active) nodes get preference in selecting slots in PACT's distributed slot assignment procedure at the start of each frame, which consists of a number of mini control slots where nodes claim their share of the following sequence of data slots.
PEDAMACSPower Efficient and Delay Aware Medium Access: A TDMA scheme that extends the common single hop TDMA to a multi-hop sensor network, using a high-powered access point to synchronize the nodes and to schedule their transmissions and receptions.Equivalent to [Arisha:2002].
PicoRadioNN: A multi-channel CDMA scheme proposed for the PicoRadio project. A second, simple (tone) radio is used to wakeup neighbors and have them switch on the full radio.
PMACPattern-MAC: A distributed, adaptive TDMA-like scheme in which nodes exchange traffic patterns to establish in which slots they should be awake to accomodate the load. Within a slot nodes use CSMA/CA to establish who will actually be sending (and receiving).
Preamble samplingAloha with Preamble Sampling: Reduces the idle listening overhead of classical ALOHA by transmitting a long reamble, allowing receivers to poll periodically for incoming traffic. Very similar to LPL, which is based on CSMA.
RATE ESTRate Estimation MAC: uses a separate wakeup radio to awake the destination when needing to send a data packet. Includes an optimization to avoid waking up ALL 1-hop neighbors. Each data packet includes an interval for the receiver to schedule a triggered wake-up on the data radio to receive the next (batch of) packet(s). The length of the interval depends on the estimated data rate of the application. The use of triggered wakeups makes RATE EST more efficient than STEM.
RI-MACReceiver-Initiated MAC: shares the LPL-philosphy of having the sender pay the price, but reverses the roles in the rendez-vous action to minimize channel usage (i.e. no long preambles). It is the receiver who sends out invitation beacons at regular intervals, where a sender must wait until it sees one and respond by sending the message. Collisions are detected at the reciver, who then sends out a new beacon specifying a contention window, increasing its length on consecutive collisions.
S-MACSensor MAC: The classic CSMA-style protocol for sensor networks. Nodes synchronize on time (by building virtual clusters) and employ a FIXED duty cycle to save energy (i.e. to reduce idle listening overhead). S-MAC includes carrier sense, collison avoidance (RTS/CTS signalling), and overhearing avoidance.
S-MAC/ALSensor MAC with adaptive listening: Reduces multi-hop latency of S-MAC by extending the active period when handling traffic to see if more is coming. Equivalent to T-MAC's adaptive duty cycling mechanism.
SCP-MACScheduled Channel Polling: Evolution of S-MAC/AL with efficient channel polling (LPL). All nodes synchronize on a common slot structure. Sending nodes contend for sending out a busy tone (step 1) just before the beginning of a slot to "wakeup" receivers. The data msg is send next (step 2) using standard contention resolution. Thus the overhead of contention resolution is largely paid by the senders (in step 1) unlike with standard LPL where all nodes listen in (only step 2).
SEESAWSEESAW: is quite similar in spirit to LPL in that nodes periodically listen for incoming traffic and senders pay the price in additional transmission time to catch the listen interval. However, listening is performed at the packet level, and each node chooses its own interval, which allows for balancing energy consumption (heavily loaded nodes use short listen intervals). A heuristic is used to automatically tune a node's individual protocol parameters (listen period and interval, number and rate of notification packets, etc.)
SiftSift: A randomized CSMA protocol that uses a fixed-length contention window with a non-uniform probability distribution of transmitting. This provides very effective collision resolution across a wide range of the number of senders.
SMACSSelf-organizing Medium Access Control for Sensor networks: uses FDMA to avoid collisions between links. Requires a regular "find neighbor phase" in which links are discovered and assigned to random frequency band. The underlying assumption that there are many bands does not hold for current low-power radios.
SS-TDMASelf-Stabilizing MAC: All trafic is scheduled in a fixed sequence of rounds (e.g., north, south, east, west) to guarantee collison-free transmissions. Very simple, but of limited applicability (GRID-like topologies only).
STEMSparse Topology and Energy Management: Uses two radios. A very low-power one to wakeup a target node. And a second, full-fledged radio for data communication. For efficiency low power listening (see LPL) is used on the signalling radio, and to reduce the associated latency target nodes explicitly acknowledge the wakeup on the signalling radio.
T-MACTimeout MAC: An improvement on basic S-MAC. To handle traffic fluctuations in time and space T-MAC uses an ADPATIVE duty cycle, implemented as a timeout after the last event Problems with high loads (early sleeping of 2nd hop neighbors) and multi-hop latency (max 3 hops per slot). Equivalent to S-MAC/AL.
TRAMATRaffic-Adaptive Medium Access: A TDMA scheme based on a distributed slot selection algorithm, which requires nodes to exchange two hop information, but only every 100 frames to amortize the overhead. When running out of packets nodes may release their slots for use by neighbors in between schedule updates.
WiseMACWireless sensor MAC: Improves on LPL by remembering the poll schedules of neighbors, which allows for sending short preambles just in time. Falls back to long preambles when clock drift gets too large.
X-MACX-MAC: addresses the overhearing overhead associated with LPL's long preambles by using a strobed sequence of short packets including the target ID allowing for fast shutdown and response. Very similar to CSMA-MPS. X-MAC also includes a lookup table to automatically select the optimal check interval based on the (estimated) traffic load.
Z-MACZebra MAC: A hybrid scheme that starts off as CSMA but switches to TDMA when the load increases. Nodes run a distributed slot selection algorithm and the owner of a slot gets precedence by using as small random back-off value. Since all nodes must listen to all slots, Z-MAC builds on LPL for energy efficiency.
[Arisha:2002]NN: A TDMA scheme where the sink (cluster head) periodically computes a schedule based on traffic and battery-level information from the nodes within reach. Direct node-to-node communication is possible.
last modified: 12-11-2010 10:06:49.0000000000 CET