CN102625426B - Double optimized topology control method of energy consumption equalization and delay in heterogeneous wireless networks - Google Patents

Abstract

The invention discloses a double optimized topology control method of energy consumption equalization and delay in heterogeneous wireless networks. The method comprises the following stages: logic neighbor selection, link symmetrization and logic neighbor adjustment. For a heterogeneous wireless network, the first two stages are executed at the beginning of generation or update of a topology; and the logic neighbor adjustment is executed between two topology updates so as to extend update cycle of the topology and reduce update times of the topology. According to the double optimized topology control method of the energy consumption equalization and the delay in the heterogeneous wireless networks, both the energy consumption equalization and the delay are considered, and optimization of the network topology is achieved.

Description

The two topology control methods of optimizing of energy consumption balance and time delay in heterogeneous wireless network

Technical field

The invention belongs to the applications of computer network and wireless network topology control technology, relate to a kind of Topology Control of wireless sensor network, particularly the two topology control methods of optimizing of energy consumption balance and time delay in a kind of heterogeneous wireless network.

Background technology

In the Internet of Things epoch, wireless sensor network link up play a part aspect physical world and information world very important.The sensor node with RFID reader functionality can read the RFID label information on the object in its wireless signal coverage in contactless mode, in the mode of multi-hop, is transferred to information centre, realizes the inquiry of object, monitoring, tracking.Wireless networking is simple, easy care, up-front investment are little, easily expansion.But node is with powered battery, and energy reserve is limited, under some deployed environment, energy cannot supplement (as the ECOLOGICAL ENVIRONMENTAL MONITORING in field), energy exhaust the termination that means network life.Even under environment that can be supplementary at energy (as the monitoring to major port open depot), also wish that energy can exhaust in certain balanced as far as possible mode, so both can reduce the stability (as alleviating the impact of possible energy cavity on network performance) that charge frequency also can maintain the network performance during twice charging interval.Existing relevant topology control technology (forms once energy consumption balance is topological as set forth the consideration of energy consumption balance not comprehensively, can how long, how maintain safeguard, the problem such as how maintenance cost and topological performance compromise), do not consider the optimization to time delay.The transmission of the events such as the leakage of stolen, the toxic gas of goods, fire alarm has harsh requirement to time delay.The required mass data transfers of the application such as video monitoring has strengthened the necessity that time delay ensures especially.

The network that heterogeneous wireless network is often referred to plurality of wireless networks technology (as 2G and 3G cell communication network, WLAN, WiMax etc.) and deposits, in the present invention, refer in particular to the network that the sensor node by parameter isomery forms, these parameters comprise node maximum transmission power, receiving sensitivity, initial energy reserve value, processor ability, storage and buffer memory ability etc.

Summary of the invention

The present invention wants technical solution problem to be to provide energy consumption balance and the two topology control methods of optimizing of time delay in a kind of heterogeneous wireless network, in this heterogeneous wireless network, the two optimization of energy consumption balance and time delay topology control method can be realized and take into account energy consumption balance and time delay, realizes the optimization to network topology.

The present invention solves the problems of the technologies described above adopted technical scheme to be:

The two topology control methods of optimizing of energy consumption balance and time delay in a kind of heterogeneous wireless network, comprise that logic neighbor choice, link symmetrization, logic neighbours adjust three phases, for a heterogeneous wireless network, at the beginning of topology generation or topology renewal, carry out the first two stage, between twice topology upgraded, actuating logic neighbours adjust to extend the topology update cycle, reduce topological update times.

Described logic neighbor choice comprises the following steps: (node is when determining transmitting power, its all nodes that cover are called its physical neighborhood, if only consider energy optimization target, reducing transmitting power can corresponding minimizing physical neighborhood number, now without using this concept of logic neighbours.When considering that a plurality of targets are carried out topological optimization, node neighbours number just can not only rely on and regulate transmitting power to control.In the present invention, each physical neighborhood covering while getting maximum transmission power for node is calculated a kind of tolerance of taking into account energy consumption balance and time delay, according to metric, select from big to small some nodes as neighbours, now node should regulate its transmitting power to cover these neighbours, but this transmitting power also may cover more node, therefore need this concept of logic neighbours to refer in particular to the neighbours of node approval.Logic neighbours are limited to a desired value, mainly consider, only allow logic neighbours participate in the forwarding of pathfinding bag, also only have selecteed logic neighbours to participate in the forwarding of route response bag and packet, therefore, logic neighbours count I to simplify Route Selection, but its quantity can affect network-in-dialing very little.)

Arbitrary node i is carried out following algorithm and is selected k arest neighbors node:

1, node i is at moment t ₁use its maximum transmission power P _{t-max, i}broadcast packet contains the packet of following nodal information: its unique identify label id _i, maximum transmission power P _{t-max, i}, the coordinate (x in Euler's plane _i, y _i), initial energy reserve E _i, at moment t ₁time dump energy

2, node i receives the packet that comprises described nodal information of its all physical neighborhood broadcast, and is stored in its neighbor list N _l(i) in;

3, node i value of being respectively calculated as follows, and be stored in its neighbor list N _l(i) in:

3.1, for collecting each physical neighborhood in N (i), neighbours calculate link metric value;

To j physical neighborhood, calculate the metric ∧ of link i → j _j;

3.2, for neighbours collect the physical neighborhood that each dump energy in N (i) is greater than average primary power, calculate the first energy consumption balance factor; The physical neighborhood that is not more than average primary power for each dump energy is calculated the second energy consumption balance factor;

The first energy consumption balance factor

The second energy consumption balance factor

for physical neighborhood j is in t dump energy constantly;

E _{av, i}average primary power for all nodes in N (i): m collects the nodes in N (i), E for neighbours _jprimary power for node j;

3.3, for neighbours collect each physical neighborhood computation delay factor of influence in N (i);

The delay factor gamma that physical neighborhood j is corresponding _{ij, del}computing formula be: t _ijrepresent that node i is to the communication delay of j

3.4, for neighbours collect in N (i) each physical neighborhood j, calculate link metric Ψ: when physical neighborhood dump energy is greater than

During average primary power, when physical neighborhood dump energy is not more than average primary power,

4, node i selects k physical neighborhood as logic neighbours by Ψ value descending order: preferential selection has physical neighborhood, if having nodes be less than k, from having node according to value fill vacancies in the proper order from big to small; If all physical neighborhood numbers are not more than k, choosing entirely;

5, node i is calculated from it to the necessary minimum emissive power of selected logic neighbours, and the logic neighbor information of these performance numbers and selection is recorded to neighbor list N _l(i) in.If (node i is to its neighbours' (as node j) distance d _ijbe less than cross distance d _crossover, the computing formula of minimum emissive power is $P_{t-\min -\mathrm{ij}}=\frac{P_{\mathrm{rx}-\mathrm{thresh}}\·{\left(4\mathrm{\π}\right)}^2{\left(d_{\mathrm{ij}}\right)}^{2}L}{G_t{G}_r{\mathrm{\λ}}^2},$ Otherwise use formula $P_{t-\min -\mathrm{ij}}=\frac{P_{\mathrm{rx}-\mathrm{thresh}}\·{\left(d_{\mathrm{ij}}\right)}^4}{G_t{G}_r{h}_t^2{h}_r^2},$ Wherein, P _rx-threshbe the neighbours' (as node j) of node i receiving sensitivity [receiving sensitivity is known, and every type of sensor node has its specific Sensitirity va1ue, and it is a specific character of equipment], L is and propagates irrelevant system loss factor, G _tand G _rbe respectively transmitter antenna gain (dBi) and receiving antenna gain, λ is carrier signal wavelength, h _tand h _rrespectively height of transmitting antenna and reception antenna height.)

Metric ∧ _jcomputational methods be: angle line between node i and base station and the angle between link i → j, if node i is not known orientation, base station, wherein, γ is path loss index, and c is by formula c=d _ij/ d _charcalculate d _charfor characteristic distance.

Formula in, wherein, t ^brepresent that node j forwards the time of the Frame that a frame length is l; for being that node j calculates SINR value [Signal Interference and Noise Ratio] by node i, wherein, p _efor ambient noise [p _eneed actual measurement to obtain, also can rule of thumb estimate-unit is watt], ρ _kjto be node k actual interference receive a kind of probability of data to node j to ∈ [0,1], i.e. probability of interference, and N (j) is interference source set, represent when node i sends to node j the signal strength signal intensity of receiving at node j place.[ receiving equipment by node j place detects, and unit is watt.】

For described probability of interference ρ _kj, it is that node i is that the neighbours of node j collect the probability of interference ρ that in N (j), each node k calculates _kj, be characterized in:

1) for ease of node i, calculate, the data that in N (j), each node k need to record self send situation, and regularly report node j, then by node j, offer node i and use;

2) node k need to record beginning and the finish time of its each data frame transfer, due to the restriction of node buffering area, As time goes on, deletes old record, only retains the state-of-the-art record suitable with buffer pool size;

3) when node k need to record to node j report data frame transfer, add that it is the delivery time of each Frame mark, after node j receives, convert at once self clock scale to;

4) when node j prepares to deliver these records and uses to node i, add the delivery time of node j self, after node i reception, also convert at once self clock scale to;

5), if Frame is shorter, ignores the sending and receiving time, otherwise can calculate according to Frame length and transmission rate;

When sending the Frame of node k time scale in record, node i converts to after self time scale, just can with the Frame transmission record of time period, compare with self, by comparison, show that intersection accounts for the percentage of total comparison part, uses this percent value as probability of interference ρ _kjtolerance, and use formula ρ _kj=α ρ _kj+ (1-α) ' ρ _{kj, new}upgrade ρ _kjvalue, wherein, α is a smoothing factor [one desirable 7/8 (7/8ths) of α], it has determined an old probability of interference ρ _kjthe shared weight of value.

Link symmetry approach is as follows:

Arbitrary node i is carried out following algorithm makes it and the backfence link of logic become symmetric links: [reason of link symmetrization is the basic demand (as needs oppositely send ACK frame) of wireless network MAC agreement, if link is asymmetric, make wireless network MAC agreement realize complicated.】

1, node i is at moment t ₂use its maximum transmission power P _{t-max, i}broadcast packet contains the packet of following information: its unique identify label id _i, neighbor list N _l(i);

2, node i receives the packet that comprises above-mentioned information of its neighbours' transmission, and calculates as follows neighbours' collection with symmetric links:

2.1, whether its each physical neighborhood of node i inspection is chosen as logic neighbours by it;

2.2, if so, but it is not considered as the other side logic neighbours, the other side added as to logic neighbours and send out confirmation to the other side; Otherwise, wait for the other side's confirmation; [effect of confirmation is, the fact that first node of action need to allow the interdependent node of rear action know to have taken action to be to avoid repetition similar operations, and the content of this confirmation comprises the type coding of initiating node identification and receiving node sign and identifying this confirmation.】

If 2.2.1 sent out confirmation to the other side, return to step 2.1;

If 2.2.2 wait for the other side confirmation and overtime before receive, return to step 2.1, otherwise by the other side from logic neighbours concentrate and delete, then return to step 2.1;

2.3 if not, return to step 2.1 and check next physical neighborhood, until all physical neighborhood are all checked through.

The method that described logic neighbours adjust is as follows:

Twice logic neighbor choice execution interval phase, arbitrary node i is carried out following logic neighbours' method of adjustment:

1, when running into following situations, node i is according to the energy consumption P of one hop link _relay(d)=(α ₁+ α ₂d ^δ) r, wherein, α ₁=α ₁₁+ α ₁₂, and α ₁₁and α ₁₂respectively reflector and receiver electronic devices and components energy consumption [α ₁₁get 26.5nJ/bit, α ₁₂get 59.1nJ/bit], α ₂without line amplifier energy consumption [α ₂get 10pJ/bit/m ²(γ=2) or 0.0013pJ/bit/m ⁴(γ=4)], d is the distance between reflector and receiver, γ is path loss index [γ can have different values, and conventionally in horizon range, γ is desirable 2, otherwise value is between 2 and 4, in some cases, even can reach 6], d ^γthe meaning be the γ power of d; From N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption: [specific energy consumption be by after the energy consumption in statistics a period of time again divided by drawing during this period of time, with variable, f represents; If consider the change histories factor of energy consumption, with another variable u, represent, and use formula u=α .u+ (1-α) f to dynamically update, wherein α

Represent the shared weight of history value.] [one desirable 7/8 (7/8ths) of α]

1.1, to certain neighbour, send or when the final destination of forwarding messages bag and message bag is not these neighbours when node i;

1.2, when node i is when receiving the message bag of certain neighbours' forwarding;

1.3, when node i listens to certain neighbours' forwarding messages bag;

2, when running into following situations, node i is according to formula P _t(D)=(α ₁₁+ α ₂.d ^γ) .r is from N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption:

2.1, when node i is when receiving the message bag of certain neighbours' transmission;

2.2, when node i listens to certain neighbours' transmission packet;

P wherein _t(d) be with speed rbits/s, to send the energy consumption of message; α ₁₁α ₂.d ^γimplication is the same;

3, when running into following situations, node i is according to formula P _t(d)=(α ₁₂) .r is from N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption:

3.1, when node i sends or when the final destination of forwarding messages bag and message bag is these neighbours to certain neighbour;

3.2, when node i listens to certain neighbours' receipt message bag;

P wherein _t(d) be with speed rbits/s, to send the energy consumption of message; α ₁₂, r implication is the same;

4, node i periodically checks N _l(i) in, whether have residual energy value to be less than and logic neighbor node (our thinking is to want to find out the replacement object that NOT logic neighbor node in the physical neighborhood of the node i that meets specified conditions is usingd as logic neighbours) that specific energy consumption is greater than NOT logic neighbours if having, after confirming, the other side replaces it (if there are a plurality of NOT logic neighbours that meet the demands, following the preferential principle of the large person of link metric Ψ value);

If 5 confirm or reached to replace the number of times upper limit without maybe obtaining the other side, actuating logic neighbor choice algorithm and link symmetrization algorithm carry out topology adjustment.[this topology adjustment is larger adjustment, larger implication refers between node needs mutual exchange message to upgrade topology, this needs larger energy consumption, and logic neighbours adjustment only need be according to existing information, owing to not carrying out information exchange, so energy consumption is less, in contrast to this, need the topology renewal of mutual exchange message to be just called larger topology adjustment.】

[above-mentioned 5 steps can only be carried out 1 time or 2 times, also can dynamic circulation carry out, the time delay of usining is greater than default threshold value as exit criteria, because the logic neighbours here adjust only demonstration, has considered the harmony of energy consumption and has shown consideration time delay, the possibility that exists time delay constantly to increase.】

It is 5 that logic neighbours count k.

Beneficial effect

In heterogeneous wireless network of the present invention, the two topology control methods of optimizing of energy consumption balance and time delay are improving on the existing basis based on energy consumption balance Topology Control deficiency, the integrated consideration to delay Optimization, propose the two topology control methods of optimizing of new energy consumption balance and time delay, to meet, have the network application of delay performance requirement needs.Meanwhile, propose to forming the controlled dynamic maintaining method of topological cost and predictable topological update method, so that the reasonable compromise of network application to Considerations such as the requirement of topological performance, topological renewal frequency, maintenance costs.

It is a kind of the active necessary technology of wireless network links number range to be set that topology is controlled, and has network-in-dialing and the effect in optimized network life-span of maintaining.By ensureing the availability of link, Topology Control also can maintain the stability of Route Selection and reduce routing cost.Feature of the present invention: (1) defines suitable topological performance metric so that select more liberally via node, to reach as far as possible the two objects of optimizing of network node energy-consumption harmony and end-to-end transfer of data time delay; (2) based on existing local knowledge, infer topological renewal opportunity, by topological neighbours, adjust and reduce topological renewal frequency to reduce topological maintenance cost.

Beneficial effect of the present invention (advantage) is summarized as follows:

1) can take into account the two optimization aim of energy consumption balance and end-to-end transmission delay;

2) energy consumption balance is better than existing correlation technique;

3) end-to-end transmission delay is better than existing correlation technique;

4), because average residual energy mean square deviation is less than existing correlation technique, so the topology update cycle can extend, be conducive to reduce topological maintenance costs;

5) because end-to-end transmission delay jitter value is less than existing correlation technique, so be conducive to ensure the transmission quality of video stream data.

Accompanying drawing explanation

Fig. 1 is a kind of variation tendency (based on jumping figure route) of single network topology lower node average residual energy mean square deviation;

Fig. 2 is average end-to-end delay variation tendency (based on jumping figure route) between a kind of single network topology lower node and base station;

Fig. 3 is a kind of variation tendency (based on Energy-aware routing) of single network topology lower node average residual energy mean square deviation;

Fig. 4 is average end-to-end delay variation tendency (based on Energy-aware routing) between a kind of single network topology lower node and base station;

Fig. 5 is the variation tendency (based on jumping figure route) of the average nodal dump energy mean square deviation of comprehensive multiple network topology;

Fig. 6 is average end-to-end delay variation tendency (based on jumping figure route) between the node of comprehensive multiple network topology and base station;

Fig. 7 is the variation tendency (based on Energy-aware routing) of the average nodal dump energy mean square deviation of comprehensive multiple network topology;

Fig. 8 is average end-to-end delay variation tendency (based on Energy-aware routing) between the node of comprehensive multiple network topology and base station;

Fig. 9 is average nodal dump energy mean square deviation comparison (based on jumping figure route) in each scheme under different node densities;

Figure 10 is average end-to-end delay comparison (based on jumping figure route) between node and base station in each scheme under different node densities;

Figure 11 is average nodal dump energy mean square deviation comparison (based on Energy-aware routing) in each scheme under different node densities;

Figure 12 is average end-to-end delay comparison (based on Energy-aware routing) between node and base station in each scheme under different node densities

Embodiment

Technical conceive and the technology path of invention are described below:

Although our method, applicable to Euler space, is lost a characteristic of stock for sake of convenience and not, we consider the flat topology network that the n of a kind of random placement in Euler's plane node forms.V represents the set of node in network, and ε represents to represent the nonoriented edge collection of communication link.Therefore, network diagram can be expressed as g=(v, ε).We also use g _d=(v, ε _d) represent the network diagram containing directed edge, wherein ε _drepresent the communication link collection that comprises directed edge.

In node set v, arbitrary node i has unique identify label id _i, by the x of Euler's plane, y coordinate is determined its position.I → j represents ε _din a directed edge, its distance table is shown d _ij.If there is nonoriented edge between node i and j, available represent,

Our setting network node is randomly dispersed in a rectangular area.The method of this deployment types of a kind of modeling is to adopt two-dimentional poisson process ^[1]. node equalization appears at finite region A Anywhere possibly, finds that the probability of n node is as follows in the A of region ^[1]:

$\mathrm{Pr}\left[\mathrm{n\; nodes\; in\; A}\right]=e^{-\mathrm{\λ}}\·\frac{{(\mathrm{\λ}\·A)}^2}{n!}---\left(1\right)$

In formula (1), λ is the Poisson process density relevant to network node density.The neighbours of node collect i can be expressed as N (i): n _l(i) be used to store the state of each node in N (i), the power that it contains node identify label, energy reserve, topological performance metric parameter, needs with each neighbours signal post.Each node has individual maximum transmission power P _t-max, and required transmitting power is expressed as P from node i to j _tij.Node i is expressed as e in the dump energy of time t _i ^t.In radio communication, transmitting power P _tx, received power P _rx, meet following relation with the parameter such as the transmitting-receiving distance d of point-to-point transmission and path loss index γ:

P _rx∝P _tx/d ^γ (2)

In formula (2), depend on that whether sighting distance (line-of-sight, LOS) link exists, γ can have different values, and conventionally in horizon range, γ is desirable 2, otherwise value is between 2 and 4, in some cases, even can reach 6.

The present invention adopts the wireless transceiver energy consumption model of using in document [2,3].This model adopts variable transmit power to meet minimum receiving sensitivity demand.Most of existing wireless transceivers are supported the variable transmit power rank of a plurality of discrete step.The energy consumption of wireless transceiver mainly consumes on the devices such as Digital Signal Processing, front-end circuit, power amplifier.The energy consumption that sends message with speed rbits/s is calculated by following formula:

P _T(d)＝(α ₁₁+α ₂.d ^γ).r (3)

The energy consumption that receives same message is calculated by following formula:

P _T(d)＝(α ₁₂).r (4)

In formula (3) and (4), α ₁₁and α ₁₂be constant, its value depends on the factors such as digital encoding and decoding mechanism, modulation and demodulation, pulse shaping filtering.α ₂depend on the factors such as antenna performance, channel condition, amplifier effect, receiver sensitivity.Two kinds of widely used radio transmission models are free space model (γ=2) and two-wire propagation model (γ=4).As for selecting which model to depend on the distance between communication node.Document [2] has defined a kind of cross distance (crossover distance) d that determines this selection _crossover, its mathematic(al) representation is as follows:

$d_{\mathrm{crossover}}=\frac{4\mathrm{\π}\sqrt{L}{h}_r{h}_t}{\mathrm{\λ}}---\left(5\right)$

If internodal distance is less than d _crossover, select free space model, otherwise select two-wire propagation model.Document [4] provides the relational expression of transmitting power in free space model, received signal strength, transmitting-receiving node spacing etc.:

$P_{\mathrm{rx}}\left(d\right)=\frac{P_{\mathrm{tx}}{G}_t{G}_r{\mathrm{\λ}}^2}{{\left(4\mathrm{\π}\right)}^2{d}^{2}L}---\left(6\right)$

In formula (5) and (6), λ is carrier signal wavelength; L is and propagates irrelevant system loss factor; h _tand h _rrespectively height of transmitting antenna and reception antenna height; G _tand G _rrespectively transmitter antenna gain (dBi) and receiving antenna gain; D is the distance between transmitting-receiving node.Between transmitting-receiving node, without the region of line of sight link, bilinear model is more accurate than free space model, and its mathematic(al) representation is as follows:

$P_{\mathrm{rx}}\left(d\right)=\frac{P_{\mathrm{tx}}{G}_t{G}_r{h}_t^2{h}_r^2}{d^4}---\left(7\right)$

Document [3] has provided the theoretical upper bound of supporting the linear topology wireless sensor network life of multi-hop communication.Its model is based on characteristic distance (characteristic distance) d _charconcept calculate optimal number of hops.This distance value is relevant with receiver hardware and channel characteristics etc. to reflector.To any transmitting node t,, there is an optimal number of hops K in receiving node r, and the distance D between them _opt, can be calculated by following formula:

Characteristic distance d _charirrelevant with distance D, can be calculated by following formula:

$d_{\mathrm{char}}=\mathrm{\γ}\sqrt{\frac{{\mathrm{\α}}_1}{{\mathrm{\α}}_2(\mathrm{\γ}-1)}}---\left(9\right)$

In formula (9), α ₁=α ₁₁+ α ₁₂.

Concerning arbitrary node i, the position of its neighbours j can be in order to control oneself as the coordinate (d in the center of circle _chara, d _charb) represent.Coordinate figure is characteristic distance d _charfunction.The distance of node i and its neighbours j can be expressed as d _ij=d _charc, wherein $c=\sqrt{a^2+b^2}.$

In data transmission procedure, the energy consumption of one hop link can be expressed as follows:

P _relay(d)＝(α ₁+α ₂d ^γ)r (10)

Node i to base station energy efficiency the available following formula of power consumption in high path calculate:

P _link-min(D)＝K _opt·P _re _lay(d _char) (11)

In fact, node i may not be along optimal path to the transfer of data of base station, may be a path that distance is D >=D, and available following formula calculates:

$P_{\mathrm{link}}\left(D^{\′}\right)={\mathrm{\Σ}}_{i=1}^K{P}_{\mathrm{relay}}(c_i\·d_{\mathrm{char}})---\left(12\right)$

Document [5] use formula (13) is below as the tolerance of the energy efficiency of the data transfer path from arbitrary node i to base station, and its value is larger, and energy efficiency is higher, and has proved satisfied formula (14) below.

In formula (14), with be respectively a and the normalized mean value of c on transmission path, meet respectively following relation:

$\tilde{a}=\frac{K_{\mathrm{opt}}}{K}---\left(15\right)$

${\stackrel{\‾}{c}}^{\mathrm{\γ}}\≤\frac{{\mathrm{\Σ}}_{i=1}^K{\left(c_i\right)}^{\mathrm{\γ}}}{K}---\left(16\right)$

In document [5], to single-hop path, use single-hop parameter value to substitute average parameter value, therefore there is following expression:

In formula (17), ∧ _jthat node i is calculated and its neighbours j between the tolerance of link, angle line between node i and base station and the angle between link i → j.If node i is not known orientation, base station, available following formula estimation:

Document [5] is by introducing γ _j=e _j(here, E is the primary power of each node to/E, e _jthe dump energy of node j) consider the harmony of energy consumption, therefore, to the tolerance of link, can be calculated by following formula:

Ψ _j＝∧ _j·γ _j (19)

When carrying out topology control, according to Ψ _jthe K of a selective value maximum neighbour.If maximum neighbours' number is not enough K, choosing entirely.But it supposes that each node has identical primary power, and this exists limitation in actual applications.We improve as follows:

Neighbours for arbitrary node i collect the arbitrary node j in N (i), and its primary power is E _j, the dump energy of t is at any one time the average primary power E of all nodes in N (i) _{av, i}can be calculated by following formula:

$E_{\mathrm{av},i}=\frac{{\mathrm{\Σ}}_{j\∈N\left(i\right)}{E}_j}{M}---\left(20\right)$

In formula (20), M collects the nodes in N (i) for neighbours.When carrying out topology control, preferentially select dump energy to be greater than the neighbours of average primary power.If this type of neighbours' number is greater than K, it is logic neighbours that the value of calculating according to formula (21) is selected a maximum K physical neighborhood.

In formula (21), by following formula, calculated:

If dump energy is greater than neighbours' number of average primary power and is less than K, and the dump energy that has sufficient amount is not more than the neighbours of average primary power, the value that can calculate according to formula (23), according to value descending order selects required logic neighbours' number to supply.If this class neighbours number is still not enough, without calculating by formula (23), and directly select all neighbours.

In formula (23), by following formula, calculated:

Above method can guarantee that neighbours that dump energy is greater than average primary power become logic neighbours' first-selection, are conducive to balanced energy consumption.

On above-mentioned improved basis, we further take into account link delay.The data frame transfer success rate formula adopting from document [6] known, the Signal Interference and Noise Ratio that link receiving terminal (being node j) is located (Signal to Interference Noise Ratio, SINR) β _ivalue larger, data frame transfer success rate is higher.The one way time delay of a common link comprises forward delay interval, the propagation delay on transmission medium of transmitting terminal (or forward end), the reception delay of receiving terminal.If energy direct communication between two nodes, for the purpose of simplification problem, communication delay is set as 0, disregards transmitting time, propagation time, time of reception.If two internodal transfer of data need to forward by intermediate node, these two internodal communication delays depend on the disposal ability sum of intermediate node.Therefore, we adopt following formula to calculate the communication delay of one hop link:

$t_{\mathrm{ij}}=\frac{t_j^b}{{(1-e^{-0.5{\mathrm{\β}}_{\mathrm{ij}}})}^l}---\left(25\right)$

In formula (25), represent that node j forwards the time of the Frame that a frame length is 1.

On link i → j, the received power value of receiving node j transmission power level with sending node i relation, according to Euclidean distance d from node i to node j _ijto be less than or greater than d _crossover, can determine that it meets free space model or two-wire propagation model.Near receiving node j, may there is other node, for example k.When node i is communicated by letter with node j, the send action of node k causes interference to node j, and the size of disturbing transmitting power with node k between relation, according to Euclidean distance d from node k to node j _kjto be less than or greater than d _crossover, can determine that it meets free space model or two-wire propagation model.When node i is communicated by letter with node j, the SINR at its communication link quality and node j place is closely related, also directly affects the Forwarding Latency of node j to Frame.We propose a kind of new method of the SINR of calculating value, and its feature is to have considered the actual probability that causes interference in potential interference source.According to the method, by node i, be that node j calculates the formula of SINR value as follows:

${\mathrm{\β}}_{\mathrm{ij}}=\frac{p_{\mathrm{ij}}^r}{p_e+{\mathrm{\Σ}}_{k\∈N\left(j\right)}{\mathrm{\ρ}}_{\mathrm{kj}}\·p_{\mathrm{jk}}^r}---\left(26\right)$

In formula (26), p _erelevant with ambient noises such as thermal noise, electromagnetic noises, ρ _kj∈ [0,1] is that node k actual interference receives a kind of probability of data to node j.For ease of calculating based on local message, only consider in set N (j) that node is as the situation of interference source here.Provide a kind of probability of interference ρ below _kjcomputational methods.

When node i is that node j calculates Signal Interference and Noise Ratio β _ijtime, need to first for collecting each node k in N (j), the neighbours of node j calculate probability of interference ρ _kj.For ease of node i, calculate, in N (j), each node k need to record the data transmission situation of self, and regularly reports node j, then provides node i to use by node j.Node k need to record beginning and the finish time of its each data frame transfer, due to the restriction of node buffering area, As time goes on, can delete old record, only retains the state-of-the-art record suitable with buffer pool size.When node k need to record to node j report data frame transfer, add that it is the delivery time of each Frame mark.After node j receives, convert at once self clock scale to.When node j prepares to deliver these records and uses to node i, add the delivery time of node j self.After node i receives, also convert at once self clock scale to.If Frame is shorter, can ignores the sending and receiving time, otherwise can calculate according to Frame length and transmission rate.Said method can adapt in network ubiquitous clock out of step conditions between each node preferably.When node i sends the Frame of node k time scale in record, convert to after self time scale, just can with the Frame of time period, send record with self and compare.By comparison, can show that intersection accounts for the percentage of total comparison part.We use this percent value as probability of interference ρ _kjtolerance, and upgrade ρ with following formula _kjvalue.

ρ _kj＝α·ρ _kj+(1-α)·ρ _kj，new (27)

In formula (27), α is a smoothing factor, and it has determined an old probability of interference ρ _kjthe shared weight of value.

For arbitrary node i, we are that arbitrary j ∈ N (i) introduces a delay factor gamma _{ij, del}, its calculating formula is as follows:

${\mathrm{\γ}}_{\mathrm{ij},\mathrm{del}}=\frac{t_{\mathrm{ij}}}{{\mathrm{\Σ}}_{k\∈N\left(i\right)}{t}_{\mathrm{ik}}}---\left(28\right)$

On the basis of formula (21), we propose to have considered as follows the relational expression of delay:

On the basis of formula (23), we propose to have considered as follows the relational expression of delay:

If a node becomes the logic neighbours of more other nodes, the bag that it need to be born forwards task just may be heavier, energy consumption also can be faster, therefore, needs cycling service topology control program to meet specific energy consumption balance requirement to maintain topological performance.The speed degree that the length of topology update cycle and topological performance worsen is closely related, but existing related work not yet has a quantitative solution to this.We propose a kind of to forming the controlled dynamic maintaining method of topological cost and predictable topological update method, and its basic ideas are as follows:

Our idea is, if can delay the process that topological performance worsens as far as possible, can extend the cycle that topology is upgraded, and saves the communication overhead of topological control information exchange and the time overhead that topological control program is carried out.Our specific practice is: in any node i, carry out after once topology is upgraded, it not only needs to remember that each becomes the related parameter values of logic neighbours' node, also will remember the related parameter values of other physical neighborhood in its maximum transmission power coverage.

According to formula (10), we know, if statistics actual traffic, just can approximate evaluation Energy Expenditure Levels.Therefore, node i need to send or forwarding data to which logic neighbour, correspondingly it just can upgrade these neighbours' dump energy, and upgrades its specific energy consumption (being used for measuring the speed degree that energy consumes).If node i is idle, need to intercept that all neighbours in its maximum transmission power coverage send or the situation of forwarding data, correspondingly upgrade their dump energy and specific energy consumption.From source or the target MAC (Media Access Control) address of Frame, can know which neighbour is sender or forwarding person.

In some cases, as when receiving-transmitting sides is all the neighbours of node i, can also listens to which neighbour and receive data.At this moment, receiving power consumption also will add up.According to these record values, can in the logic neighbours of node i, occur dump energy lower than and specific energy consumption during higher than the node of other the arbitrary physical neighborhood in the maximum transmission power coverage of node i, consider to carry out logic neighbours' replacement.When obtaining selected physical neighborhood agreement, can replace.If cannot replace, logic neighbours' number of node i will be tending towards reducing (or exhausting rapidly because of some node energy, or because being deleted from the logic neighbor list of node i, these nodes can not get new supplementing), when lower than certain value, will affect network connectivty, therefore, need to restart topological refresh routine.

Owing to only considering that energy consumption index replaces, after repeatedly replacing, it is unacceptable that time delay index may become, and at least probability of interference is (as ρ _kjvalue) update status fail to be reflected in time during topology adjusts, therefore, even in the situation that replacing smoothly, after replacing for 1～2 time, also need to restart topological refresh routine.

Although said method extended and to have saved the time overhead that topological control program is carried out because of the topology update cycle, also introduced the time overhead that the dump energy of node and specific energy consumption are upgraded.Even so, the communication overhead that extends the topological control information exchange of saving because of the topology update cycle is a kind of net earnings, and this net earnings is embodied directly in the lifting of network application data throughput.

Our topology control method comprises that logic neighbor choice, link symmetrization, logic neighbours adjust three phases.At the beginning of topology generation or topology renewal, carry out the first two stage.Between twice topology upgraded, actuating logic neighbours adjustment algorithm is to extend the topology update cycle, to reduce topological update times.Only have logic neighbours to participate in the forwarding of pathfinding bag, also only have selecteed logic neighbours to participate in the forwarding of route response bag and packet, therefore, logic neighbours count I to simplify Route Selection, but its quantity can affect network-in-dialing very little.Document [5] provides conclusion by theory analysis and emulation, at network node, follows in Poisson distribution situation, and when logic neighbours count k>5, network-in-dialing probability can not increase again.Therefore, we also count k by logic neighbours and are decided to be 5.By the following formula in document [7], can calculate and in the network design district of two-dimentional Poisson distribution, guarantee that network node 1-is communicated with the deployment density λ that probability is Pr.

$\mathrm{Pr}\≈{(1-e^{-\mathrm{\λ}\·\mathrm{\π}\·d_{\max }^2})}^n---\left(31\right)$

In formula (31), d _maxbe the emission maximum radius of node, n is the quantity of nodes.But this formula is the situation for the emission maximum radius isomorphism of node, when the emission maximum radius isomery of node, can use the mean value of each node emission maximum radius to replace as approximate.

Below in conjunction with the drawings and specific embodiments, the invention will be further described.

Embodiment 1:

This method both can be used for the environment (as being deployed in the wireless sensor node that region, major port open depot doubles as RFID label reader) of the controlled deployment of network node, also can be used for the environment (as the ECOLOGICAL ENVIRONMENTAL MONITORING in field) of the uncontrollable deployment of network node.For the latter, we first consider that 100 node deployments are at 500 * 500m ²plane domain on network (i.e. this Area Node density of forming individual/m ², Poisson distribution is obeyed in the position of each node), then consider in same plane domain, nodes to be increased to respectively 200 and the situation of 300 o'clock.In addition, dispose a sink node (being base station) in the upper right corner in this region.

Base station noenergy restriction, the initial energy reserve of other node has randomness, but maximum is no more than 0.5J.For not making the too complicated and randomness that can pay close attention to the emphatically initial energy reserve impact on different topology control method of problem, our other parameter of setting network node is all identical, as shown in table 1.The reason that the initial energy reserve of consideration node has randomness is: no matter adopt what measure, the lack of uniformity of energy consumption exists all the time, therefore, when initial, in appointed area, shed at random some nodes, can be dead a part of after use a period of time, so supplement.Network condition after several times can be used as a new network and treats so repeatedly, has formed the network that a kind of node initial energy reserve has randomness.

Table 1 network node parameter list

In our method, if each node (as node i) calculate and its neighbours (as node j) between considered angle during the tolerance of link (being line between node i and base station and the angle between link i → j), is called scheme 1, otherwise is called scheme 2.Equally, in the method for the document [5] comparing with our method, if considered angle be called scheme 3, otherwise be called scheme 4.

In emulation, we are random selects 30 nodes to base station, to send data (data transfer rates is in Table 1), these nodes adopt following two kinds of metric forms when carrying out Route Selection: a kind of is to adopt jumping figure, the another kind of mean value that adopts both link ends residue energy of node.Route based on the former is called shortest path route, and route based on the latter is called Energy-aware routing.100 time steps of our emulation, in each time step, 30 nodes choosing at random respectively send the data of a designated length.For scheme 1 and 2, every 5 time steps carry out logic neighbours to be adjusted, and for scheme 3 and 4, does not have logic neighbours to adjust function.

Performance index are relatively network node average residual energy mean square deviation and the average end-to-end transmission delay of network node.Network node average residual energy mean square deviation refers to that the residual energy value of each node in network and the sum of square of deviations of network node average residual energy value get average evolution again.The average end-to-end transmission delay of network node refers to the average of end-to-end transfer of data time delay between the node of all transmission data and base station.

Fig. 1, Fig. 3, Fig. 5, Fig. 7 have shown the variation tendency of average residual energy mean square deviation with simulation time.From these figure, our scheme one and two average residual energy mean square deviation are all less than existing scheme, i.e. scheme three and four.This illustrates that the energy consumption balance of our scheme is better than existing scheme.Because Fig. 1 and Fig. 3 are for single topology situation, therefore, the energy consumption balance sex expression meeting between different topology is variant, but general trend is consistent.

What Fig. 5 and Fig. 7 showed is the mean value of 10 kinds of energy consumption balance tolerance under network topology, and wherein the data under each topology are the mean value of its 100 time step simulation results, and therefrom visible, the advantage of our scheme is obvious and stable.From these figure, we also see, when network node adopts Energy-aware routing, the dump energy mean square deviation of scheme one is milder with the variation of simulation time, this is because Energy-aware routing is always selected the more link of dump energy, thereby is more conducive to energy balance than the route based on jumping figure.

From Fig. 2, Fig. 4, Fig. 6, Fig. 8, can see, the average end-to-end transmission delay of network node shows that with the variation tendency of simulation time our scheme one and two has advantage equally.The variation tendency that Fig. 2 shows is more regular, Fig. 4 has shown a kind of randomness, this is because the data transfer path based on jumping figure Route Selection that Fig. 2 adopts is more stable, and otherness between each time step is is mostly that total length by the data transfer path of each time step causes.Because Fig. 4 adopts Energy-aware routing routing, along with the differentiation that each node energy consumes, data transfer path is also difficult to stablize, thereby shows a kind of randomness on end-to-end transmission delay.

Fig. 6 and 8 is the mean value of data under 10 kinds of network topologies, and wherein the data under each topology are the mean value of its 100 time step simulation results.As can be seen from Fig. 6, except some shake of emulation initial stage, demonstrate subsequently a kind of very stable state, its reason is mainly also that the data transfer path bringing based on jumping figure route is comparatively stable.

As seen from Figure 8, scheme one, two and three average end-to-end transmission delay are milder with the variation of simulation time, and scheme four fluctuations are larger, this is mainly not carry out neighbor choice because scheme four is not considered time delay factor when carrying out topology control, adopts Energy-aware routing routing to cause data transfer path to be difficult to stablize in addition.Although scheme three is not considered time delay factor yet, the azimuth between its consideration and base station, has limited it and has selected neighbours' scope, and it is so frequent that the change of data transfer path does not have scheme four, thereby time delay fluctuation is also be not as large as scheme four.

Fig. 9 and 11 shows the average residual energy mean square deviation contrast situation of each scheme under heterogeneous networks node density.No matter scheme three and four adopts jumping figure route or Energy-aware routing, variation tendency is consistent (along with the increase of node density, average residual energy mean square deviation reduces), this is mainly that it only considers energy consumption and energy consumption balance selection neighbours, when network node density is large, choice is also large, therefore, is conducive to energy balance.

The average residual energy mean square deviation of our scheme has obvious consistency variation tendency with the variation of network node density like that not as scheme three and four, for example, in Fig. 9, network node density increase is conducive to balance energy consumption generally, but increases to certain value to no effect.This is mainly that our scheme has been considered time delay factor, and time delay is also relevant to interference, network node density increases the interference that may cause and also increases, therefore, in network node density, increase to after certain value, its positive benefit of bringing will be offset by negative results.

Figure 11 shows, under compared with macroreticular node density, our scheme energy consumption balance effect is not as existing scheme, and this is to be mainly conducive under the environment of energy balance at Energy-aware routing, and our scheme has retrained its range of choice to logic neighbours owing to having taken into account to the consideration of time delay.But under lower network node density, our scheme has clear superiority.In practical application, network node was disposed secret meeting increases Financial cost, thereby highdensity network is unrealistic, even for fault-tolerant, disposed redundant node, can take the method for dormancy in turn too, the node density of simultaneously working still can be too not high, therefore can not affect the practical application effect of our scheme.

Figure 10 and 12 has shown to consistency the impact of network node density on end-to-end delay, and node density is higher larger to delay.On the whole, our scheme is conducive to reduce time delay, especially in the situation that node density is less.In Figure 10, our scheme two shows poor under nodes higher density, this is mainly that nodes higher density has increased its selection and deviates from the possibility of base station neighbors, once what Figure 10 adopted selectes path, more difficult change based on jumping figure route simultaneously, therefore, the impact in unfavorable path is more difficult disappears at once.In Figure 12, due to what adopt, be based on Energy-aware routing, therefore, under this environment, there is not above-mentioned situation in our scheme two.

Above emulation shows, in the situation that network node deployment density is not too high, our scheme all has advantage in the network topology performances such as energy consumption balance and end-to-end transmission delay.This advantage comes from our scheme to node initial energy reserve otherness, innovation modeling based on factors such as physical disturbance model lower node Forwarding Latencies, and twice network topology reproducting periods inferred and adjudicated the Creative Design of logic neighbours being adjusted etc. to strategy based on existing local knowledge in front and back.

List of references

[1]Chao，X.，Dargie，W.，Lin，G.，2008.Energ _y model for H2S monitoring wireless sensor network.In：CSE‘08：Proceedings of the200811th IEEE International Conference on Computational Science and Engineering.IEEE Computer Seciety，Washington，DC，USA，pp.402-409.

[2]Heinzelman，W.B.，2000.Application-specific protocol architectures for wireless networks.PhD Thesis.(Supervisor-Chandrakasan，Anantha P.and Supervisor-Balakri shnan，Hari)

[3]Timothy，M.B.，6arnett，T.，Ch，A.P.，2001.Upper bounds on the lifetime of sensor networks.In：ICC2001，pp.785-790.

[4]Rappaport，T.，2001.Wireless Communications：Principles and Practice.Prentice Hall PTR，Upper Saddle River，NJ，USA.

[5]Waltenegus Dargie，Rami Mochaourab，Alexander Schi l l，Lin Guan.A topology control protocol based on el igibi l ity and efficiency metrics.The Journal of Systems and Software，2011，84：2-11.

[6]G.Bacci，M.Luise，and H.V.Poor，Game-theoretic power control in impulse radio UWB wireless networks，ArXiv e-prints，2007.

[7]P.Santi.Topology control in wireless ad hoc and sensor networks.ACM Computing Surveys，2005，37(2)：164-194.

Claims (6)

1. in a heterogeneous wireless network, topology control methods are optimized in energy consumption balance and time delay pair, it is characterized in that, comprise that logic neighbor choice, link symmetrization and logic neighbours adjust three phases, for a heterogeneous wireless network, at the beginning of topology generation or topology renewal, carry out the first two stage, between twice topology upgraded, actuating logic neighbours adjust to extend the topology update cycle and reduce topological update times;

Described logic neighbor choice comprises the following steps:

Arbitrary node i is carried out following algorithm and is selected k arest neighbors node:

1), node i is at moment t ₁use its maximum transmission power P _{t-max, i}broadcast packet contains the packet of following nodal information: its unique identify label id _i, maximum transmission power P _{t-max, i}, the coordinate (x in Euler's plane _i, y _i), initial energy reserve E _iwith at moment t ₁time dump energy

2), node i receives the packet that comprises described nodal information of its all physical neighborhood broadcast, and is stored in its neighbor list N _l(i) in;

3), the node i value of being respectively calculated as follows, and be stored in its neighbor list N _l(i) in:

3.1, for collecting each physical neighborhood in N (i), neighbours calculate link metric value;

To j physical neighborhood, calculate the metric Λ of link i → j _j;

3.2, for neighbours collect the physical neighborhood that each dump energy in N (i) is greater than average primary power, calculate the first energy consumption balance factor; The physical neighborhood that is not more than average primary power for each dump energy is calculated the second energy consumption balance factor;

The first energy consumption balance factor

The second energy consumption balance factor wherein j represents j physical neighborhood;

for physical neighborhood j is in t dump energy constantly;

E _{av, i}average primary power for all nodes in N (i): , M collects the nodes in N (i), E for neighbours _jprimary power for node j;

3.3, for neighbours collect each physical neighborhood computation delay factor of influence in N (i);

The delay factor that physical neighborhood j is corresponding computing formula be: t _ijrepresent that node i is to the communication delay of j;

3.4, for collecting each physical neighborhood j in N (i), neighbours calculate link metric Ψ: when physical neighborhood dump energy is greater than average primary power, when physical neighborhood dump energy is not more than average primary power,

4), node i selects k physical neighborhood as logic neighbours by Ψ value descending order: preferential selection has physical neighborhood, if having nodes be less than k, from having node according to value fill vacancies in the proper order from big to small; If all physical neighborhood numbers are not more than k, choosing entirely;

5), node i is calculated from it to the necessary minimum emissive power of selected logic neighbours, and the logic neighbor information of these performance numbers and selection is recorded to neighbor list N _l(i) in;

Metric Λ _jcomputational methods be: angle line between node i and base station and the angle between link i → j, if node i is not known orientation, base station, wherein, γ is path loss index, and c is by formula c=d _ij/ d _charcalculate d _charfor characteristic distance.

2. the two topology control methods of optimizing of energy consumption balance and time delay in heterogeneous wireless network according to claim 1, is characterized in that formula in, wherein, t ^brepresent that node j forwards the time of the Frame that a frame length is l; for being that node j calculates SINR value by node i, wherein, p _efor ambient noise, ρ _kjto be node k actual interference receive a kind of probability of data to node j to ∈ [0,1], i.e. probability of interference, and N (j) is interference source set, represent when node i sends to node j the signal strength signal intensity of receiving at node j place.

3. the two topology control methods of optimizing of energy consumption balance and time delay in heterogeneous wireless network according to claim 2, is characterized in that, for described probability of interference ρ _kj, it is that node i is that the neighbours of node j collect the probability of interference ρ that in N (j), each node k calculates _kj, be characterized in:

1) for ease of node i, calculate, the data that in N (j), each node k need to record self send situation, and regularly report node j, then by node j, offer node i and use;

2) node k need to record beginning and the finish time of its each data frame transfer, due to the restriction of node buffering area, As time goes on, deletes old record, only retains the state-of-the-art record suitable with buffer pool size;

3) when node k need to record to node j report data frame transfer, add that it is the delivery time of each Frame mark, after node j receives, convert at once self clock scale to;

4) when node j prepares to deliver these records and uses to node i, add the delivery time of node j self, after node i reception, also convert at once self clock scale to;

5), if Frame is shorter, ignores the sending and receiving time, otherwise can calculate according to Frame length and transmission rate;

When sending the Frame of node k time scale in record, node i converts to after self time scale, just can with the Frame transmission record of time period, compare with self, by comparison, show that intersection accounts for the percentage of total comparison part, uses this percent value as probability of interference ρ _kjtolerance, and use formula ρ _kj=α ρ _kj+ (1-α) ρ _{kj, new}upgrade ρ _kjvalue, wherein, α is a smoothing factor, it has determined an old probability of interference ρ _kjthe shared weight of value.

4. the two topology control methods of optimizing of energy consumption balance and time delay in heterogeneous wireless network according to claim 1, is characterized in that, link symmetry approach is as follows:

Arbitrary node i is carried out following algorithm makes it and the backfence link of logic become symmetric links:

1), node i is at moment t ₂use its maximum transmission power P _{t-max, i}broadcast packet contains the packet of following information: its unique identify label id _iwith neighbor list N _l(i);

2), node i receives the packet that comprises above-mentioned information of its neighbours' transmission, and calculates as follows neighbours' collection with symmetric links:

2.1, whether its each physical neighborhood of node i inspection is chosen as logic neighbours by it;

2.2, if so, but it is not considered as the other side logic neighbours, the other side added as to logic neighbours and send out confirmation to the other side; Otherwise, wait for the other side's confirmation;

If 2.2.1 sent out confirmation to the other side, return to step 2.1;

If 2.2.2 wait for the other side confirmation and overtime before receive, return to step 2.1, otherwise by the other side from logic neighbours concentrate and delete, then return to step 2.1;

2.3 if not, return to step 2.1 and check next physical neighborhood, until all physical neighborhood are all checked through.

5. the two topology control methods of optimizing of energy consumption balance and time delay in heterogeneous wireless network according to claim 1, is characterized in that, the method that described logic neighbours adjust is as follows:

Twice logic neighbor choice execution interval phase, arbitrary node i is carried out following logic neighbours' method of adjustment:

1), when running into following situations, node i is according to the energy consumption P of one hop link _relay(d)=(α ₁+ α ₂d ^γ) r, wherein, α ₁=α ₁₁+ α ₁₂, and α ₁₁and α ₁₂respectively reflector and receiver electronic devices and components energy consumption, α ₂be without line amplifier energy consumption, d is the distance between reflector and receiver, and γ is path loss index, d ^γthe meaning be the γ power of d; From N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption:

1.1, to certain neighbour, send or when the final destination of forwarding messages bag and message bag is not these neighbours when node i;

1.2, when node i is when receiving the message bag of certain neighbours' forwarding;

1.3, when node i listens to certain neighbours' forwarding messages bag;

2), when running into following situations, node i is according to formula P _t(d)=(α ₁₁+ α ₂d ^γ) r is from N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption:

2.1, when node i is when receiving the message bag of certain neighbours' transmission;

2.2, when node i listens to certain neighbours' transmission packet;

P wherein _t(d) be with speed r bits/s, to send the energy consumption of message; α ₁₁α ₂d ^γimplication is the same;

3), when running into following situations, node i is according to formula P _t(d)=(α ₁₂) r is from N _l(i) in this neighbours' residual energy value of preserving in, deduct corresponding power consumption values, and upgrade its specific energy consumption:

3.1, when node i sends or when the final destination of forwarding messages bag and message bag is these neighbours to certain neighbour;

3.2, when node i listens to certain neighbours' receipt message bag;

P wherein _t(d) be with speed r bits/s, to send the energy consumption of message; α ₁₂, r implication is the same;

4), node i periodically checks N _l(i) in, whether there is residual energy value to be less than and specific energy consumption is greater than NOT logic neighbours' logic neighbor node, if having, after the other side confirms, replace it;

5) if confirm or reached to replace the number of times upper limit without maybe obtaining the other side, actuating logic neighbor choice algorithm and link symmetrization algorithm carry out topology adjustment.

6. according to the two topology control methods of optimizing of energy consumption balance and time delay in the heterogeneous wireless network described in claim 1-5 any one, it is characterized in that, it is 5 that logic neighbours count k.