Method and system of communications
TECHNICAL FIELD OF THE INVENTION
The present invention relates to communications. More es¬ pecially it relates to multiple access communications over channels of diverse channel qualities, e.g. signal to noise and interference ratios, varying over time. Particularly it relates to traffic distribution and channel allocation for efficient communications over such channels.
BACKGROUND AND DESCRIPTION OF RELATED ART
Multiple Access Communications are previously known. In e.g. mobile communications systems such as GSM (Global Sys¬ tem for Mobile Communications) • or UMTS (Universal Mobile Telecommunications System) users or user equipment are al¬ located communications resources depending on demand and availability.
Multiple Access Communications usually relies upon multi¬ plexing technologies for dividing or splitting a channel resource into components of more limited capacity. Exam¬ ples of such technologies are FDM (Frequency Divisions MuI- tiplex) , TDM (Time Division Multiplex) and CDM (Code Divi¬ sion Multiplex) with associated multiple access technolo¬ gies FDMA (Frequency Division Multiple Access) , TDMA (Time Division Multiple Access) and CDMA (Code Division Multiple Access) respectively. According to prior art, users are multiplexed by dividing an entire bandwidth resource into channels or channel resources characterized by orthogonal¬ ity in frequency, time and code domain, respectively. Also known in prior art are multiplexing systems combining two or more of FDM, TDM and CDM thereby achieving channels or channel resources characterized by orthogonality in two or more domains, e.g. time and frequency domain.
Prior art recognizes both, circuit switched communications and packet switched communications. In circuit switched communications, communications resources are allocated, even if they temporarily or due to channel conditions could be released, for an entire communications session, e.g. an entire phone call or an entire telephone modem data connec¬ tion of one or more data transfers. In packet switched communications, communications resources are allocated ac¬ cording to communications requirements for distribution of packet not necessarily forming an entire communications session, e.g. a fraction of a digitized spoken sentence or a fraction of a data file.
From prior art channel probing is also previously known. When an HF (High Frequency) channel is probed for optimum frequency of operation or a radio communications channel is probed for appropriate transmission power level, it is an example of channel probing.
Syad Faraz Shamim, 'How does Unequal cost Path Load Balanc¬ ing (Variance) Work in IGRP and EIGRP?, ' CISCO July 31, 2003, describes a routing protocol in a fixed network that allows random based forwarding to one of several one routers.
R. Nelson and L. Kleinrock, 'The spatial Capacity of a slotted ALOHA multihop packet radio network with capture, ' Trans. On Comm. , June 1984, also describes random based forwarding of packets to one out of several packet radio network routers, but in addition ensures that a packet is heading in a generally correct direction.
J. Jubin and J. D. Tornow, 'The DARPA packet radio network protocols, ' IEEE Proceedings, pp. 21-32, Jan. 1987, de¬ scribes alternate path routing, allowing a packet which is retransmitted over a link to be duplicated while multicast
to several nodes from which the packet again follows a shortest path routing approach.
International Patent Applications WO96019887 and WO98056140 exploit the opportunistic idea to enhance system capacity of a cellular system or at a base station.
U.S. Patent Application US2002/0183086 discloses channel probing in a CDMA system. A mobile station increases transmission power level until a base station acknowledge¬ ment is received. The power level at which an acknowledge- ment is received is stored and forms a basis for the power level at which a second probe is initiated thereby reducing time to acknowledgement of second probe.
U.S. Patent US6546045 reveals channel probing in a communi¬ cations system for selection of one out of two available modulation schemes. A probe signal is transmitted for es¬ timation of channel multipath and doppler characteristics. Adaptive modems measure the communication channel's doppler and multipath characteristics. Upon the occurrence of pre¬ determined criteria, the channel scattering function esti- mate may be updated and a new modulation scheme may be se¬ lected to continue transmission.
U.S. Patent US6400783 reveals channel probing in a communi- nications system for channel estimation and channel equali¬ zation.
M. W. Subbarao and B. L. Hughes, Optimum Transmission Ranges and Code Rates for Frequency-Hop Packet Radio Net¬ works' IEEE Transactions on Comm. r April 2000, describes optimization of information forward progress (or informa¬ tion efficiency forward progress) . Optimal transmission range and code rates are investigated in a highly loaded
frequency hopping packet radio network by an information efficiency forward progress performance measure.
None of the cited documents above discloses multi-user di¬ versity forwarding or at a transmitter forwarding of data, data and receiver of the forwarding being dependent on channel quality to various receivers.
SUMMARY OF THE INVENTION
A general problem of routing or relaying of information from a sender to a receiver over intermediary nodes for a scarce medium, such as forwarding of radio channels, is making efficient use of the scarce resource thereby provid¬ ing for a great traffic capacity.
State of the art radio network routing or relaying offer limited spectrum, efficiency and data rates more limited than necessary not making full use of diversity effect of a plurality of relaying nodes or full use of the fluctuations of channel for transmissions scheduling.
An effect of this is state of the art requiring excessive transmission power and providing inferior quality of ser- vice compared to what is achievable.
Consequently, there is a need of optimizing channel use and data scheduling for a packet data network forwarding data over a plurality of relay nodes, particularly when there is data for a plurality of destinations.
It is consequently an object of the present invention to achieve a communications system maximizing an objective function reflecting network performance.
A further object is to achieve increased throughput or re¬ duced delay and to achieve a system of increased traffic
capacity at a given performance, e.g. in terms of through¬ put and delay.
It is also an object increase quality of service in a multi-user system.
Another object is to reduce peak power and energy consump¬ tion for a given performance level.
Finally, it is an object to categorize users perceiving good and bad propagation properties respectively and allo¬ cating and multiplexing users accordingly.
These objects are met by a method and system of multi-user diversity forwarding opportunistically optimizing perform-r . -.•.!•■ ance over available relay nodes and data packets , buffered.•'-•.,..i for forwarding. ' < r I ■ > •♦,
BRIEF DESCRIPTION OF THE DRAWINGS ,< , ...rV ',,
Figure 1 illustrates a preferred mode of the invention,',';.'.,?-.,•, where a four phase process is completed during a single time slot «TS n» for a sequence of time slots «TS n-2», «TS n-l», «TS n», «TS n+l», «TS n+2».
Figure 2 illustrates a simplified flowchart for a method according to the invention.
Figure 3 illustrates example information forward-progress versus hop distance for various transmission powers, ac¬ cording to the invention.
Figure 4 illustrates example information forward-progress for a rayleigh fading channel with transmission power P equal to 1 W and for various constants C2 corresponding to various node densities, according to the invention.
Figure 5 illustrates a block diagram of a forwarding node according to the invention.
Figure 6 illustrates schematically a cut from a communica¬ tions system according to the invention comprising a plu- rality of forwarding nodes within transmission range from a forwarding node.
DESCRIPTION OF PREFERRED EMBODIMENTS
According to the invention communications are divided into four phases: 1. inquiry phase,
: .2. response phase, . ■;-
3. data phase, and
4. data acknowledge, ACK, phase. '■■■.■
The four phases are described in detail below. • ., o.
Figure 1 illustrates a preferred mode of the invention, where all four phases are completed during a single time slot «TS n» for a sequence of time slots «TS n-2», «TS n- 1», «TS n», «TS n+l», «TS n+2».
For a communications network comprising a plurality of com- munications nodes, if it has been determined that node i should transmit in timeslot n, transmission power P1 is de¬ termined for subsequent transmissions from node /. Trans¬ mission power can be set to be allowed to change between transmissions. Preferably, the transmission power Pi should reflect network topology changes and depend on transmit buffer fill level, previously failed transmissions and Quality of Service, QoS related parameters.
The invention also allows for node transmission power adap¬ tation depending on whether a small power consumption or a
maximum communications performance is preferred for a par¬ ticular node.
The decision to transmit is preconditioned on packets wait¬ ing in the transmit buffer and on the medium access princi- pie adopted, e.g. Slotted ALOHA with randomly selected transmission time-instances.
1. The inquiry phase can adopt different methods, wherein the first method is based on that each transmitting station sending an inquiry message with a locally unique word (used for correlation by the receiver) at transmit power P1. A receiving node j may then identify the sender of the inquiry message and the power level at which it was received., .<■
In a second mode of the invention the transmitter address is included in inquiry messages, which are coordinated; not to collide or interfere. This is preferably achieved through support of a collision-free protocol, known in the art. Further the inquiry message preferably 'includes transmission power level Pi, to be used in subsequent trans¬ missions and optionally also receiver station identity for one or more desired receiver stations.
2. In the response phase, each station sends a response message comprising information on actual or expected signal to noise ratio, SNR, or signal to interference and noise ratio, SINR, to the transmitter accompanied by identity of the transmitter. As an alternative each station determines which rate can be supported for reception, as channel fre¬ quency selectivity can easily be incorporated into the de¬ cision, and respond with the supported rate in place of SNR. The rate can be an explicit data rate or implicitly indicate the rate by, e.g., particular modulation and for¬ ward error correction. A further alternative is to return
the channel estimate enabling rate selection at transmitter side.
For all three alternatives, the response is preferably sent in accordance with a collision-free protocol.
Based on channel quality, in terms of SNR, supported rate, channel frequency selectivity or channel estimate, a selec¬ tion is made at the transmitter. In a network or system of a plurality of transmitter the selection is preferably made jointly for the transmitters. The decision comprises two or three issues:
1. Selection of which out of a plurality of data packets residing in buffer to transmit.
2. Selection of receiver of the data packet.
3. Selection of transmission parameters, e.g. modulation and forward error control code. in a preferred mode, also more precise channel state information is included as selection pa¬ rameter for determining of appropriate trans¬ mission parameters. The channel state informa- tion is either included in the response message from the receiver or estimated at the transmit¬ ter assuming channel reciprocity, when appro¬ priate.
The third issue is not included for embodiments not allow- ing rate adaptation.
3. In the data phase, selected packet is sent to the deter¬ mined receive node with the selected transmit parameters.
4. In the ACK phase, the receiving node responds with an acknowledgement indicating whether the packet was received correctly or not.
Figure 2 illustrates a simplified flowchart for a method according to the invention. The principles of ARQ (Auto¬ matic Repeat Request) are considered well known in the art and are not described in further detail. The method is re- iterated «Tl», «T10», «R1», «R8» for each transmission in¬ terval determines whether there are data in transmission buffer «T2». If so, in a first step of selection transmis¬ sion power is selected «T3». An inquiry message or probe is sent «T4» from the transmitter and is received in one or more receivers «R2». The receivers determine SNR and op¬ tionally other parameters as described above «R3». A re¬ sponse message comprising relevant parameters is sent back «R4» from each receiver to the transmitter, where it is re¬ ceived «T5». The transmitter compares received response messages and their parameters for determining of packet to send, receiver to forward packet to and communications link parameters «T6». The data packet is transmitted, option¬ ally with an accompanying list of one or more specified re¬ ceivers according to preference as described above. The data packet is sent «T7» and received at one or more re¬ ceivers «R5» and performs an error detection of the re¬ ceived packet. If the received packet is in accordance with an error detecting code an acknowledgement is sent «R6». When the acknowledgement is received in the trans- mitter «T8», appropriate actions are undertaken at the transmitter, e.g. deleting the packet from the transmission buffer if the ARQ protocol is a stop-and-wait protocol or a selective repeat protocol. If the error detection code de¬ tects errors in the received packet no positive acknowledg- ment is sent to the transmitter from the receiver «R7». Depending on ARQ protocol used also erroneous packets may be useful when combined with subsequent transmissions of identical data (Hybrid-ARQ) , and therefore stored «R7». If
the packet is correctly received, it is forwarded to the user application in accordance with the ARQ protocol «R7».
When the packet has been transmitted, the process stops «T1O», «R8» and reiterates «Tl», «R1» as long as there are data in the transmission buffer.
According to the invention an opportunistic selection is made among relay nodes for an opportunistically selected particular data packet, in the sense that the data packet is forwarded to one or more relay nodes for which the transmission resource is made best use of according to a specified measure. This can include, e.g., forward pro¬ gress, data rate, channel quality or quality of service, as well as fairness between users and nodes.
For the routing there is also an. underlying cost minimizing or metric maximizing protocol,, with associated algorithm, in addition to the proposed forwarding scheme. The cost- minimizing algorithm is preferably a well-known algorithm, such as the Bellman Ford algorithm using average rate, en¬ ergy, delay or. hop cost/metric. Preferably, the cost- minimizing algorithm operates on a larger time scale than the path/link variations, and hence gives an average indi¬ cation of where to route data. Additionally, the route de¬ termination protocol or a topology control protocol pro¬ vides the basic setting of transmission power of data sent.
In the forwarding process the derived path/link cost or metric for a destination is taken into account, and pref¬ erably also faster varying characteristics of links from the forwarding node under consideration. In addition to packet and route selection, link performance of a particu- lar link is preferably optimized.
The optimization is rendered possible by the inquiry- response phase and the selection among multiple users to send data to.
According to the invention an example metric maximizing rate times traversed distance is adopted.
As an illustrative example, consider an infinite number of relay nodes positioned along a line in the desired direc¬ tion of routing, searching for the optimum node to forward a data packet to and the data rate to transmit at. Infor- mation forward progress Z7 given by Shannon's formula for channel capacity times hop distance to the node to forward to then yields,
where P is transmission power, N is noise power, α is propagation constant, which equals 2 for free space propa¬ gation and for typical areas is in the range between 2 and 4, Co is a constant and B is bandwidth.
Figure 3 illustrates the information forward progress ver¬ sus hop distance for various transmission powers for exam- pie bandwidth 5=20 MHz, α=2.6 and N=kTBNF', where JVF=IO dB, 7=295 K, Ar=I.38-10"23 and demonstrates the advantage of proper path/link selection in relation to rate and hop dis¬ tance according to the invention.
The optimum distance ROpt can be upper bounded
B -ln(2) '
The invention does not require great signal constellations. As an example, with α=2 a signal constellation providing
3 bits/s/Hz would be greater than maximum achievable rate per bandwidth of α/ln(2)«l.4-oc.
In a further mode of the invention a destination dependent information progress measure is adopted. For a destination D, and a probed hop from node i to j, the destination de¬ pendent information progress is
/(Costjj ) is an objective function of Cost? , the cost of for¬ warding a data packet heading for destination D from node i to hop node j, SNRj is SNR at node j, and rate(SNRj) is the maximum rate that is supported for the signal to noise ra¬ tio at node j. The path selection for data forwarding may be determined (as indicated above) through a separate pro-, tocol with a cost-minimizing algorithm. Examples of such an algorithm are well known algorithms like the Bellman Ford and the Dijkstra algorithms. The invention is also useful with other selection algorithms and methods of com¬ bining average costs and instantaneous link conditions, e.g. instant rate.
Consequently, the data transmission buffer is scanned and destination dependent forward progress is determined for various candidate hop nodes for packets of each destination occurring in transmission buffer.
To further illustrate the benefit of the invention, an es- timation of performance improvement is given below. In rayleigh fading environment information forward progress can be determined as
Z(β,γo,R)=R-#log2(l+β-γ0)
is the information forward progress and
is the probability density function, pdf, for selection di¬ versity with, diversity order M and
γ»<*>=!§
is the average SNR as a function of distance R and
M=C2-R2
is the average number of candidate nodes on a circle with radius R around transmitting node for C2 a constant. The diversity order increases dramatically for large distances, but is also dependent on the node density, reflected by constant C2. It is noted that the fading and the presence of a large number of receiving stations are advantageous for the invention.
Figure 4 illustrates information forward progress for a rayleigh fading channel with transmission power P equal to 1 W and for various constants C2.
The invention can be combined with congestion control and e.g. adaptively vary transmission probability. Also a plu¬ rality of packets may be transmitted for each iteration in figure 2 if transmission intervals are sufficiently long for more than one packet.
When applied to a system operating according to the princi¬ ples of Orthogonal Frequency Division Multiplex, OFDM, an opportunistic selection of sub-carriers is preferably in¬ cluded. Then in a multiple access system of OFDMA (Or- thogonal Frequency Division Multiple Access) user specific data to different users is preferably multiplexed on single OFDM symbols for maximum capacity.
The careful selection of relay node and related channel properties data rate can be kept at a maximum and transmis- sion errors be kept at a minimum. A minimum interference or noise level at the relay node and a maximum signal level of received signal due to proper scheduling substantially can decrease required transmission power and requirements on received symbol energy.
The process of selection among data packets to transmit for maximum forward progress minimizes jitter and delay and in¬ creases quality of service considering all users of the communications system.
Channel probing and scheduling of the various users provide for, at each instant, pre-transmission information on transmission conditions for the various users (assuming the channel is sufficiently slowly varying) for which reasons the greater number of users and their data provide for an increased chance of finding good or excellent performance conditions to some relay node of interest.
Figure 5 illustrates a block diagram of a forwarding node «FwdNde» according to the invention. The forwarding node «FwdNde» comprises processing means «μ», receiver «Re- ceiver», reception buffer «Rx buffer», transmission buffer «Tx buffer» and transmitter «Transmitter». The reception buffer «Rx buffer» and the transmission buffer «Tx buffer» can alternatively be instrumented with one single buffer.
The processing means are informed of buffer content «RxBμ», «TxBμ». The processing means «μ» further schedules trans¬ missions in accordance with an objective criterion as de¬ scribed above depending on buffer content and nodes to for- ward to from available information on network topology.
Processing means «μ» and transmitter are also vital for channel probing, probing candidate communications links for forwarding. The transmitter «Transmitter» then sends a probe message as initialized by processing means «μ». Al- ternatively, communications link status is determined rely¬ ing on channel reciprocity as described above.
Depending on channel status the processing means. «μ>> . sets signal constellation, forwarding error correcting code or preferably both signal constellation and forward error. ;cor- recting code. The settings are communicated to the .trans-" , mitter «Txμ», «TxBμ» on transmitting side and sent «St» to receiving side. On receiving side, an inquiry message of- link probing is received «Sr» by receiver and transferred to the processing means «Rxμ», «RxBμ» for evaluation of commu- nications link status and processed for inclusion in a re¬ sponse message, the response message comprising channel status and optionally other communication parameters as de¬ scribed above, for sending «TxBμ», «Txμ», «St» to the trans¬ mitting side. Scheduled data packets are transferred from the buffer «Rx buffer», «TX buffer» in due time for trans¬ mission «St» by transmitter «Transmitter» on transmitting side. On receiving side, received signal «Sr» is detected and fed to the buffer «Rx buffer». Received data is checked for errors by processing means «μ» and an acknowl- edgment message generated if no errors are detected in re¬ ceived message. The acknowledgement message is transferred
«TxBμ», «Txμ» to the transmitter «Transmitter» and sent «St» to the transmitting side.
Figure 6 illustrates schematically a cut from a communica¬ tions system comprising a plurality of forwarding nodes ac- cording to figure 5 within transmission range from a for¬ warding node «T». Consider a forwarding situation when the center node «T» is the node on transmitting side. The net¬ work topology illustrates three candidate forwarding nodes «B», «C», «D» for the receiving side that would provide a positive forward progress for data packets heading east (right) «Destination east» and three forwarding nodes «D», «E», «P», which would provide a positive forward progress for data packets heading south-west (down-left) . «Destina- tion south-west».
The invention is not intended to be limited only to the em¬ bodiments described in detail above. Changes and modifica¬ tions may be made without departing from the invention. It covers all modifications within the scope of the following claims.