JP2008148299A - Method for route selection in cooperative relay network - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To select an optimal route in an ad-hoc network. <P>SOLUTION: The method selects a route in a wireless cooperative relay network of nodes, the nodes including a source, a set of relays, and a destination. A codeword is encoded as a data stream. The data stream is transmitted from a source to a destination via a set of relays. Mutual information is accumulated at a particular node to decode the data stream and recover the codeword. Then, a route from the source to the destination is selected based on channel state information between the nodes while transmitting and accumulating. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates generally to wireless networks, and more particularly to transmitting information in a collaborative network.

Ad hoc networks Ad hoc wireless relay networks have many advantages over traditional structured networks such as cellular networks. Ad hoc wireless networks do not require a fixed infrastructure. This can reduce costs. Since the signal can be transmitted via different routes, the reliability is increased. Energy consumption can be reduced because the intermediate relay node can receive and retransmit signals over shorter hops. Since the total transmission energy is reduced, the lifetime of the network is increased, interference is reduced, and spectral efficiency is improved.

  As a disadvantage of the ad hoc relay network, it is necessary to specify a route from the source node to the destination node. This route specifies a sequence of nodes involved in relaying information, a transmission protocol, an energy allocation at the node, and a transmission schedule.

  In a conventional ad hoc relay network, a transmission source node normally encodes information with an error correction code, and this information is transmitted via a route node sequence. The encoded information is called a codeword. As used herein, a codeword can be a file or block of data bytes.

Rateless codes In situations where the channel state information (CSI) is unknown, weak block error correction codes can make communications unreliable due to noise, while strong block error correction codes waste energy. . Rateless codes solve this problem. Instead of encoding information as a predetermined number of bits using a block error correction code, the information is encoded into a potentially infinite bit stream. The receiver then only decodes the bits of the flow until the information is restored.

  Rateless codes include fountain codes, punctured low rate block codes, Reed-Solomon codes, convolutional codes, and turbo codes. Rateless codes were originally designed for erasure channels. However, rateless codes also work well for additive white Gaussian noise (AWGN) channels. Rateless codes have also been used for Ethernet applications, point-to-point applications, broadcast applications, and multicast applications.

Routing protocol A relay node receives a signal from a previous node. The node can amplify the signal using the amplify-and-forward protocol, or the node can decode the signal information using the decode-and-forward protocol, and re-decode the information. It is also possible to encode and then retransmit the signal to the next node. This process continues until a signal is received by the destination node. Such a relay system is known as multi-hop routing. Conventionally, at any point in time, only one node forwards the signal to the next node using a single route having multiple consecutive hops from the source node to the destination node.

  A number of methods are known for selecting an optimal route in a multi-hop ad hoc relay network. In one method called “flooding”, each node broadcasts information along with the node's address and the energy (or power) used for transmission. The destination node can then select an optimal route from the accumulated amount of information received by itself. Nodes participating in the route can then be notified of the route using the feedback information.

  If all channel state information is available at a single location, the optimal route can be determined using optimization techniques or other optimization criteria that minimize the total energy consumption in the network. Distributed methods of determining the optimal route are also known.

  Another challenge for ad hoc networks is flow control. In flow control, the source node sends a continuous bit stream (flow) and the relay node directs the flow simultaneously on several parallel links to the destination node. The relay node can receive multiple such flows and can redistribute the flow to subsequent relays in the network until the flow reaches the destination node.

Cooperative Relay Network Cooperative relay networks provide an alternative to single-route multihopping. In a cooperative relay network, a plurality of nodes can transmit information simultaneously. In a conventional (modern) cooperative relay network, multiple nodes transmit the same information, and the receiving node accumulates energy from all of the signals before decoding the information. This can increase the diversity order at the receiving node. In addition, information can be sent simultaneously on several parallel routes to obtain route diversity. Therefore, the probability that the fade becomes deep is reduced. Furthermore, the energy efficiency of transmission increases due to energy integration.

  Networks using coordinated communication have a different set of problems such as parallel route selection and code selection for specific links. Expected energy costs and network topology under the assumption of cooperation are also challenges.

  Prior art route selection and optimization techniques generally assume that nodes along the route perform energy accumulation. However, the route selected in that way becomes sub-optimal when used in a network where nodes accumulate mutual information. Therefore, the prior art routing method is not applicable to the network according to the present invention.

  The routing protocol described in “Cooperative Relay Networks using Rateless Codes” of patent application 11 / 377,711, filed March 16, 2006, by Molisch et al. Assume that any node with sufficient mutual information participates in the transmission. In the semi-synchronous protocol, a specific scheduling scheme is assumed and only two hops are allowed between the source and the destination. Clearly, a two-hop network is relatively restrictive.

  Asynchronous schemes make special assumptions about the start time and duration of a node's transmission and allocate transmission power equally.

  It would be desirable to improve route selection in a more general way in networks without size constraints.

  Embodiments of the present invention provide a method for communicating information in a wireless cooperative relay network in which a receiving node can accumulate mutual information from two or more transmitting nodes. More specifically, multiple relay nodes send different encoded versions of information. The receiver can combine the different versions of mutual information during decoding. The present invention provides a method for selecting a route, transmission power, and transmission time in a network having such nodes.

  Embodiments of the present invention can optimize performance in terms of transmission time, network lifetime, energy consumption, throughput, signal to interference ratio, or other suitable criteria, or combinations thereof. Embodiments achieve this optimization by dynamically adjusting routing, transmission power, scheduling, transmission time, and used codes according to the network topology and the state of the transmission channel between nodes.

  This optimization involves performing a brute-force numerical search, a centralized suboptimal iterative search, or a decentralized suboptimal iterative search for the optimal combination route. Can be used.

  1 and 2 show cooperative relay networks 100 and 200 according to an embodiment of the present invention. The cooperative relay network 100 includes a transmission source node 111, a set of N relay nodes 121 to 124, and a destination node 131, that is, a transmission source, a relay, and a destination. It should also be understood that this embodiment of the present invention can be used with multiple relay "hops" as shown in FIG. In this case, the relay 125 also operates as a “source” for the “intermediate” relay 126. This scheme can be extended to additional hops.

  In some cases, network performance is optimized in terms of transmission time, network lifetime, energy consumption, throughput, signal-to-noise ratio, or other suitable criteria, or combinations thereof, according to constraints such as outage It is hoped that Here, the communication interruption is defined as a transmission not successful within a predetermined time or within a predetermined energy.

The source node transmits the codeword to the destination node via a set of N relay nodes. These nodes can form a single route or multiple parallel routes. The codeword can be, for example, a file of data bytes (information). The relay node uses a decoding transfer technique. The codeword has a bandwidth-normalized entropy of size H _target , given in scaled nuts / Hz (nats / Hz). Here, one nut is the amount of information in natural units, which is Euler's constant 2.718.

  The source, as well as the relay, encodes the codeword using a rateless code. All nodes operate in half-duplex mode. That is, the node can either transmit or receive, but not both at the same time. However, it should be noted that other (block) error correction codes can also be used for encoding. In the case of generalized block codes, the accumulated amount of information includes the redundancy built in those codes.

In the following, it is assumed that transmission by the direct sequence spread spectrum technique is possible. Such an approach is useful for relay networks because spreading allows transmission of different data streams in a flexible and decentralized manner, and these data streams are distinguishable at the receiver. Transmit power of node _{n i} is _{P i.} The propagation channel between nodes is modeled as a frequency flat block fading channel. The channel gain is an independent exponential distribution corresponding to the Rayleigh fading of the amplitude. Again, these assumptions are made only to provide an example, and different access schemes that may vary depending on the remaining battery life, average channel conditions, or instantaneous channel conditions. It is further noted that different transmit powers of different nodes can be used with the method.

  This receiver is a receiver that accumulates mutual information. These receivers can distinguish different information streams from different relays that may arrive at the same time. For example, the receiver can distinguish between different rateless codes using multi-user detection. In a CDMA system, when the data stream arrives with a relative delay longer than the chip duration, the receiver can process the received data streams from different relay nodes.

  The receiver includes a rateless code decoder so that the data streams from the different relay nodes can be distinguished and the mutual information of the signals transmitted by the relay nodes can be integrated. Again, it is emphasized that conventional block error correction codes can also be used.

  In one embodiment of the invention for a CDMA system, different nodes use different spreading codes in order for the destination to separate different data streams and extract the information contained in those data streams. However, different spreading codes are not always necessary. For example, the same spreading code can be used when all relays use the same code, ie rateless code or block code, to send their signals synchronously to the destination.

Pseudo Synchronous Transmission As shown in FIG. 3, codeword 301 is encoded (310) using rateless code 311 and optionally spreading code 312 to generate data stream 320. Codeword 301 can be any arbitrary collection of bits, or “information”. The source (S) 325 broadcasts the data stream received by the destination (D) 327 to a set of relay nodes (R) 326 (330). The broadcast of the data stream can be in the form of a packet.

  As shown in FIG. 4, as soon as the relay node accumulates enough mutual information to decode the data stream and recover the codeword, the relay acknowledges that the reception was successful (ACK). 400 is returned to the sender.

  The transmission source ends the broadcast after receiving a predetermined number of confirmation responses 400 and, in some cases, notifies the relay node of the end of the broadcast from the transmission source.

  At the same time, the relay node re-encodes the recovered codeword into a data stream and switches from reception to transmission.

  During this second phase, two cases can be considered. The first is when all relay nodes transmit a data stream 410 encoded with the same rateless code. The rateless code can be the same as that used by the source. Due to the delay differences inherent in transmissions from randomly placed relay nodes, the data stream reaches its destination with slightly different delays. In the following, it is assumed that these delays are longer than the chip duration but much shorter than the symbol duration. This assumption can be achieved in a direct sequence CDMA system with a large spreading factor. The destination can accumulate the mutual information amount of the data stream received from the relay node.

  The second is when each relay node uses a different rateless code and a different spreading code for transmission of the data stream 410. In this case, the destination can distinguish signals from different relay nodes through different spreading codes of those different relay nodes and accumulates mutual information. It should be noted that the destination can also use a data stream 309 broadcast by the source.

  In any case, as shown in FIG. 5, as soon as the destination successfully decodes the data stream sent by the relay and possibly the source and restores the codeword 301, it sends the signal 500 to the relay node. In some cases, broadcasting is also performed to the transmission source, and transmission is terminated.

Intuitively, binary signaling using the difference, two relays on erasure channel (erasure channel) with the loss probability p _e between the use or using multiple rateless single rateless The simplest example can be most easily understood. If those relays use the same rateless code, each bit will be lost with probability p _e ² so that 1-p _e ² bits are effectively received per relay transmission.

On the other hand, if two different rateless codes are used, the transmission is independent and 2 (1- _pe ) bits are received for each relay transmission. Note that the receiver complexity required for the two protocols is not significantly different. If the sampling rate of the analog / digital converter (ADC) of the receiver is the same as the symbol rate, both receivers require L correlators. In the first case, all correlators form a correlation with the same spreading sequence and add the results according to maximum ratio combining.

  Alternatively, the receiver can use only a single correlator. The correlator output is sampled L times during each symbol duration. This saves some hardware complexity. However, signals can arrive at irregular intervals from different relay nodes, and thus may require the ability of the ADC to sample at the chip rate. This fast sampling of the ADC greatly increases power (and therefore energy) consumption and may therefore be undesirable for battery powered sensor network applications.

  In the second case, each correlator is used to detect signals from different relay nodes. The main difference is in the decoder. The decoder becomes more complex when multiple rateless codes are used. The second method uses up multiple spreading codes to transmit one codeword, so the improved coding gain partially offsets this effect, but the spectral efficiency of the second method is worse.

  Note also that combinations of the first and second cases can be used.

Asynchronous transmission In pseudo-synchronous transmission as described above, the relay node receives only the data stream broadcast by the source node. However, when using rateless codes, the relay node can also “help” other relays to decode the data stream faster, thus accelerating the transmission process and thus energy. Consumption can be reduced. The idea is that as soon as the relay receives a sufficient amount of information to decode the codeword, the relay node starts sending the re-encoded data stream to the destination. This transmission can also be received by other relay nodes that are still in receive mode. Thus, a relay node that is in transmission mode helps a relay node that is still receiving to decode the data stream faster. That is, other relays decode both the broadcast data stream and the re-encoded data stream.

  As shown in FIG. 3, the source node starts by broadcasting the data stream to all of the relay nodes using the assigned spreading code and rateless code.

  All relay nodes continuously accumulate mutual information. When transmission by a given spreading code is started, the transmission source signals to avoid unnecessary reception of noise by the receiver. The transmission source stops transmission as described with reference to FIG.

  As shown in FIG. 6, as soon as the relay node has enough information to decode the codeword, it switches from the reception mode to the transmission mode. The relay retransmits the codeword as a data stream 601 using the spreading code and rateless code assigned to it. A relay node in receive mode observes all spreading codes simultaneously, so it can receive a data stream 330 from the source node and a data stream 602 from the relay node that is transmitting to the destination, The mutual information amount from the nodes is accumulated, and the data stream is decoded at a higher speed.

  The destination node continuously receives all the data streams of the different relays and thus accumulates the amount of information from the various relay nodes. As described above, the destination can also use a contribution 309 received directly from the source.

  As soon as the destination successfully decodes the source data stream and recovers the codeword, it broadcasts the signal 500 to the relay node and possibly to the source as shown in FIG. .

  Assume that transmission continues until the destination successfully recovers the codeword. Some energy can be saved if the source monitors the relay nodes and stops transmitting as soon as all relay nodes are in transmit mode.

Route Detection The above protocol works by implicitly determining which nodes participate in transmission, their power and transmission time. However, a process equivalent to explicitly optimizing those parameters, ie finding the optimal route (including the parameters of the nodes on the route) may be advantageous. The following describes how to find this optimal route in a network that is unlimited in its size or number of hops.

Assumptions Optimization of routes in a cooperative network is particularly difficult because more free parameters need to be optimized. Therefore, some simplifying assumptions are made, recognizing that a certain amount of overhead transmission is unavoidable in the network.

  (I) The nodes of the network are geographically randomly distributed. It is possible but not necessary that the location of some or all nodes is known through GPS or ultra-wideband positioning. Such side information can be used by the routing process as it helps provide information in the direction in which the information should be transferred. In the following, such side information is not considered. However, it is part of this patent to use such side information for routing in a collaborative network that accumulates information.

  (Ii) Route selection is centralized for simplicity. However, distributed routing can also be used, especially in large networks.

  (Iii) Instantaneous CSI or average CSI is also known in the middle. The amount of channel state information can vary. Consider a situation in which either instantaneous CSI or average CSI is usually known. CSI is due to noise and interference under observation, and / or quantization when CSI is transmitted through the network, and / or channel changes due to delays between CSI measurement and its use in the routing process. , May have noise. In the following description, only complete instantaneous CSI or complete average CSI is considered. However, other forms of CSI are included as part of this description. The cost of acquiring and distributing CSI can be incorporated into the total energy cost.

  (Iv) Assume that each node fully accumulates mutual information from different transmitters. An approximation of this assumption can be realized when rateless codes are used, where different rateless codes use different time-frequency spreading codes so that the receiver distinguishes the codes. Can do. Further, it is assumed that a rateless code of a finite size source word does not have ideal characteristics. For example, the overhead may be greater than the Shannon limit. In the following, the non-ideality of rateless codes is ignored. It is also possible for one node to bypass the transport block size limit using multiple spreading codes and / or fountain codes.

  (V) Unless explicitly stated otherwise, all spreading codes in the network are completely orthogonal, so that transmission of bitstreams or flows that use one rateless code will be transmitted with other rate codes. Assume that it does not interfere with the transmission of the encoded bitstream.

  (Vi) It is also assumed that each node can receive an arbitrarily low signal-to-noise signal, and that no overhead is required for synchronization and channel estimation at the receiver. This is an idealized assumption in the presence of constant synchronization and channel estimation overhead.

  (Vii) Information feedback 400 that the node has successfully decoded the information is assumed to be instantaneous and does not require any overhead. However, the techniques described herein are also applicable when this finite overhead and delayed transmission of feedback information is used.

  (Viii) Transmission can be unicast when a single node transmits information to a single receiving node. The transmission can also be generalized to include multiple, possibly correlated nodes (multiple flows) and / or multiple destination nodes (multicast).

  (Ix) It is assumed that the channel does not exhibit delay dispersion and can be characterized by a single scalar complex attenuation coefficient at any point in time. However, the possibility of fading (time variation) of this attenuation is also taken into account. If there is a delay spread, the spreading code is used to reduce the effect.

  (X) Assume that the source node starts transmitting a new codeword only after the destination node has successfully decoded the previous codeword. Thus, at any point in time, there is only one message from the source node in the network.

Optimization Model Conventional ad hoc routing uses the following free parameters: cooperating nodes, transmit power at each node, and the order in which the nodes transmit.

The cooperative relay network also determines the following parameters for each node i of the network: rateless code C _i (t), spreading code, transmission start time t _i , and transmission duration τ.

  The goal is to optimize any combination of the following criteria: total energy consumption of the network, total transmission time from the source node to the destination node, the spectral efficiency of transmission, and the lifetime of the network.

  As one useful measure of spectral efficiency, we define a “blocking area” around the transmitting node (although other measures of spectral efficiency can be used in the method as well). In the blocking area, radiation from the node is so strong that other communications cannot be received. Strictly speaking, the size of the blocking area depends on the detection threshold of the receiver and the propagation state in the environment. “Blocking time” is when the node is transmitting. The product of the blocking area and the blocking time can be a measure of spectral efficiency. The total number of spreading codes required to transmit a codeword is yet another parameter that characterizes spectral efficiency.

  Assume that the amount of energy available to a node is finite. That is, the node is assumed to be battery operated. The lifetime of the network “expires” when the first node uses all of its available energy and stops working. Another definition of the lifetime of the network is “time when there is no possible route from the source to the destination after that time” or “time required for the network to divide”. Network partitioning is an extensive topology change that separates the network into completely disconnected parts. This can cause sudden and serious disruption to ongoing network routing and higher layer applications.

  The optimization criterion can be in the form of a linear or non-linear function, or it can be in the form of a strict constraint. Examples of optimization criteria include, for example, a penalty function that uses a weighted sum of codeword transmission time and transmission energy, or a strict constraint on transmission time τ.

Accumulated mutual information amount Under the above assumption, the accumulated mutual information amount can be described as follows. The source node n ₀ through the N-number of the relay nodes n _i and transmits the codeword to the destination node n _{N + 1.} Each node starts transmission at time t _i with power P _i and spreading code C _i for duration τ _i . The spreading code C _i has a spreading gain SF. The instruction function I _i (t) is 1 when t _i <t <t _i + τ _i , and is 0 otherwise. The power of the received signal must be greater than the minimum threshold power ε. Furthermore, the transfer function h _ij is the complex amplitude baseband channel gain from node i to node j. In addition, the codeword has entropy H _target normalized to unit bandwidth. In that case, the mutual information amount accumulated in the node i at time t is

It becomes. Here, N ₀ is noise power. This formulation implicitly assumes that residual interference acts like noise and is not suppressed by any interference suppression mechanism such as receiver coupling detection.

If interference suppression is used, the denominator in the above equation is replaced by interference plus noise power at the output of the joint detection receiver. In order to reduce the complexity of the problem, assume that the transmission power of the node is constant. _Further, defined as _{t min,} 0 = 0, and defines implicitly _{t min, i} by 0 <i ≦ N + 1 case of _{H i (t min, i)} = H target.

A constraint condition of t ≧ t _{min, i} is imposed. In other words, any node can only start transmitting after it has decoded the codeword. Furthermore, it makes sense to constrain the transmission time to be t _i + τ _i ≦ t _{min, N + 1} , so that the node does not transmit after the codeword is decoded.

If it is desired to consider the practical requirements for signal acquisition, the constraint P _i ≧ ε _i is provided. Each node also has a maximum power constraint P _i ≦ P _max due to hardware limitations. The total energy consumed by the node n _i,

And the total energy in the network is

It is. E _i also includes the energy required for reception, as well as the energy required to obtain the channel state information and forward the channel state information to an appropriate recipient such as a central node. Note that you can.

The blocking area for a given spreading code C _m is denoted by A _m (t). From this A _m (t), a measure of the spectral efficiency S can be derived.

It is assumed that the penalty function D (E ₀ ,..., E _{N + 1} , t _{min, N + 1} , S) is minimized. Here, the function D can be a linear or non-linear function of parameters and / or can have strict constraints. For example, the service quality constraint condition may be Pr (t _{min, N + 1} > t _admissible ) <outage. Here, t _admissible is an allowable time for the message to travel from the source node to the destination node, and outage is an allowable communication interruption probability.

General Optimization Procedure The above description provides a theoretical formulation of the optimization problem. The solution of the unknown parameter can be obtained by any conventional standard optimization method such as simulated annealing, genetic algorithms, Monte-Carlo simulations. This optimization can be done at a central location or in a distributed fashion.

  FIG. 7 shows an example of optimization that minimizes the transmission time of the codeword according to the constraint on the total energy consumed for transmission.

  Given a network of nodes 701, source nodes 702, and destination nodes 703, the optimization process begins by selecting 710 an optimal non-cooperative route 711. Here, each node decodes the codeword as soon as the node accumulates a sufficient amount of mutual information. The optimal non-cooperative route can be selected according to conventional techniques.

  In order to simplify the search for the optimal route, 1 / N of the total allowable energy is first assigned to each node (720). This automatically satisfies the total energy constraint.

  Starting from the optimal uncoordinated route with energy allocation, an annealing method can be used to find nodes and power settings that reduce the total transmission time. Adjust the power setting and transmission time so that the next node along the route can still decode the codeword.

  The annealing process can operate as follows. A plurality of pairs of nodes 731 are selected at random (730). The energy of these selected pairs is adjusted as follows (740). Increase the energy assigned to one of the nodes by a small amount Δ and decrease the energy of the other node by the same amount.

  For the pair of nodes 741 whose energy has been adjusted, the feasibility of the power setting and transmission time setting, that is, the setting that satisfies the above-described constraint conditions is determined (750). Next, the power setting is selected and updated (760). This setting can be chosen randomly, or alternatively, a feasible setting that minimizes the total transmission time of the codeword from the source node to the destination node. If the resulting transmission time is less than the previous power allocation transmission time, the power settings are updated appropriately. As a result, if the transmission time becomes long, the update is performed with a certain probability p that decreases with the transmission time penalty, which leads to a decrease in performance and avoids any local minima during optimization. Even update the power.

  The optimization process is repeated (770) until convergence or a certain maximum number of iteration steps have been performed.

Specific optimization procedures Describe a number of special cases. These serve as examples of embodiments of the invention, but should not be construed as limiting the general scope of the invention.

Flooding
During the flooding technique, each node begins transmitting information at the maximum allowable power P _max at time t = t _{min, i} as soon as the node decodes the codeword. The node continues to transmit until the destination node decodes the codeword. This minimizes the total transmission time. Any transmission by a relay node assumes that the suppression of interference from this transmission at other nodes is complete, even if the distance from the relay to the destination is greater than the distance from the source to the destination. It should be noted that this leads to a reduction in the total transmission time. Intuitively, transmissions by such relay nodes help to speed up transmissions by nodes closer to the destination, thus allowing receivers of those transmissions to begin transmission earlier, thereby , The total transmission time is reduced. However, this wastes energy and even spectrum resources.

One way to improve spectral efficiency is to reuse the spreading codes of nodes that are well separated from each other. Given two nodes i and j, the transfer function h _ik from node k to node i exceeds a predetermined threshold, ie node i is “large” in the mutual information received by node k. Determine whether to contribute. For all nodes k, if the transfer function h _jk from node j to node k is smaller than the factor Z, which is the required signal-to-interference ratio, nodes i and j may use the same spreading code. it can.

Decrement Node Deduction In this approach, network flooding is first used to select routes and power allocation. Next, for each node, it is determined to reduce power consumption and increase transmission time when that node is deducted from the network. If the increase in transmission time is less than the threshold, the node is deducted. Similarly, if the increase in total transmission time due to deduction is small, nodes that increase interference due to the blocking area and blocking time of the node can also be deducted.

Incremental node addition This approach starts with the node in the best non-cooperative route as described above. For each potential node in the network, if the node is also included in the route in a coordinated fashion, the reduction in total transmission time for a given energy consumption and a given transmission time is determined separately. The node with the greatest decrease is added to the route. This procedure is repeated until the addition of another node gives only a small improvement in performance, or until the number of participating nodes reaches a predetermined threshold.

Decentralized termination In the decentralized approach, each node starts transmitting after it decodes the codeword. Transmission stops after k nodes “closer” to the destination node decode the codeword.

  Although the invention has been described by way of examples of preferred embodiments, it is to be understood that various other adaptations and modifications can be made within the spirit and scope of the invention. Accordingly, it is the object of the appended claims to cover all such variations and modifications that fall within the true spirit and scope of the present invention.

1 is a diagram of a cooperative relay network according to an embodiment of the present invention. FIG. 1 is a diagram of a cooperative relay network according to an embodiment of the present invention. FIG. FIG. 4 is a diagram illustrating a protocol broadcast step according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a protocol acknowledgment step according to an embodiment of the present invention. FIG. 6 is a diagram illustrating a protocol termination step according to an embodiment of the present invention. FIG. 5 is a diagram illustrating a synchronization protocol transmission step according to an embodiment of the present invention. FIG. 6 is a block diagram of transmission time optimization according to an embodiment of the present invention.

Claims (26)

A route selection method in a wireless cooperative relay network comprising nodes, the node comprising a source, a set of relays, and a destination, the method comprising:
Encoding the codeword as a data stream;
Transmitting the data stream via the set of relays from the source to the destination;
Accumulating mutual information at a specific node, decoding the data stream, and restoring the codeword;
Selecting a route from the transmission source to the destination based on channel state information between the nodes during transmission and integration. Route selection method.   The method of claim 1, wherein the code is a block error correction code.   The method of claim 1, wherein the code is a rateless code.   The method of claim 1, further comprising encoding the codeword using a spreading code.   The method of claim 1, wherein the codeword is an arbitrarily accumulated bit.   The method of claim 1, wherein the data stream is broadcast as packets.   The method of claim 1, wherein the route identifies nodes in the route and specifies the power, transmission time, and spreading code used by each of the identified nodes.   The method of claim 7, wherein the specifying is based on the channel state information between the nodes.   The method of claim 1, wherein the channel state information is known at a central location and the route is selected at the central location. The route identifies a node in the route, and the method includes:
The method of claim 1, further comprising: broadcasting the route to all nodes in the network.   The method of claim 1, wherein the channel state information is known in all subsets of the nodes, and the channel state information is determined in a distributed format.   The method of claim 1, wherein the selecting step uses an optimization process and optimization criteria.   The method of claim 12, wherein the optimization criterion minimizes the probability that the transmission time of the data stream exceeds a predetermined threshold.   The method of claim 12, wherein the optimization criterion minimizes energy consumption in the network.   The method of claim 12, wherein the optimization criterion minimizes maximum energy consumption at all of the nodes.   The method of claim 12, wherein the optimization criterion maximizes data throughput in the network.   The method of claim 12, wherein the optimization criterion maximizes the lifetime of the network.   The method of claim 12, wherein the optimization criterion is a linear combination of transmission time, energy consumption, throughput, and lifetime of the network.   The method of claim 12, wherein the optimization criterion is a non-linear combination of transmission time, energy consumption, throughput, and lifetime of the network.   The method of claim 12, wherein the optimization process minimizes a penalty function.   The method of claim 12, wherein the optimization process is approximate and uses a numerical optimization method.   The method of claim 12, wherein the optimization process uses an annealing method.   The method of claim 12, wherein the optimization process uses a genetic algorithm.   21. The method of claim 20, wherein the optimization process repeatedly adds nodes to select a sub-optimal route that leads to maximum improvement of the penalty function.   The method of claim 20, wherein the optimization process iteratively subtracts nodes from a sub-optimal route that leads to a maximum improvement of the penalty function.   26. The method of claim 25, wherein the optimization process is iteratively subtracting nodes from a predetermined route that do not substantially increase the amount of accumulated mutual information.

Images