An orchestration of different powered-wire and wireless technologies for multichannel Cooperative Multiple Input Multiple Output (Co-MIMO) array mesh orchestration, at the leading edge of communities in need of better or more competitive infrastructure.
This exploits the physics of light that dictate at similar given power levels, radios (or emitters and detectors for light outside of human-visible range) that transmit and sense at lower frequencies (or wider wavelengths) convey lower data bandwidth, but penetrate farther across more mediums with less attenuation. Conversely, radios at higher frequencies (or smaller wavelengths) convey much higher data bandwidth, but attenuate in all mediums (including "free" air) at much closer distances. Similarly, phased antenna arrays and virtual antennae require wider spacing when coordinating wider wavelength signals, and beamforming antenna arrays at small wavelengths can be packed into much smaller spaces. Additionally, wires that convey electrical power can also act as waveguides or direct conductors for broad bands of high data bandwidth capability, such that both data and power can be shared between any generation and storage nodes on the same wired mesh, where shorter wires can convey more data and power with less loss than longer wires.
At the software layer, this exploits recent technology developments in mesh routing and Co-MIMO multi-radio orchestration, utilizing hardware backends with new Software Defined x technologies (or SDx, where x is any technology that used to be the sole province of custom hardware and ASIC designs). In the near term, SDR (radio) and SDN (network) technologies are most useful for novel Co-MIMO mesh SDA (antenna) coordination. Developments in power microgrid controllers will similarly create SDP (power) systems which can be exploited for microgrid connections between residential renewable power and storage systems.
For a more concrete example hybrid SDP-SDN-SDR-SDA mesh with multichannel Co-MIMO orchestration, let's define some nodes with existing power wire connections and ISM/unlicensed band sharing technologies:
-
Solar and small-wind DC generation nodes, optimally with DC battery storage, connected via <=10Gbps and <=90W USB-C and 60GHz 802.11ad WiGig at <10m spacings between nodes across rooftops in grid-alike mesh topologies. These roof mounted nodes may also act as gateways into point-to-multipoint WISP or aerial wire backhauls.
-
Aggregated DC to 125VAC power inverter nodes, optionally with AC battery storage (eg. Tesla Powerwalls) or backup low-carbon generators (eg. biodiesel or wind tower turbines), connected via <=10Gbps ethernet, Power over Ethernet (PoE), ~6GHz 802.11ac MU-MIMO WiFi, and HomePlug AV2 2Gbps MIMO 20A powerlines at <=100m spacing between nodes in garages and basements. Each power and data line may convey up to 2.5KW and 10Gbps maximum, or 100W and 1Gbps at the low end (LTPoE++). HomePlug AV2 2Gbps 2.5KW connections between nodes could stretch as long as 300m. To simplify bidirectional power transmission between battery caches, any 2 nodes can run at least 2 HomePlug AV2 power lines, one for each power source to sink direction, for a total of 4Gbps and 2.5KW transit capacity in each direction. These nodes may also act as gateways to older or longer-distance infrastructure like 100KW's grids and 100km's <=100Gbps fiber.
-
~5.8GHz 802.11ac MU-MIMO WiFi nodes receiving wired power and data from one of the above node types, and orchestrating Co-MIMO radio mesh arrays at about 17m spaced increments. Due to the prevalence of tri-band 802.11ac/n WiFi COTS routers on the market that can run OpenWRT/LEDE distributions, this node type will likely be combined with the above and below node types.
-
~2.4GHz 802.11n MU-MIMO WiFi nodes receiving wired power and data from one of the above node types, and orchestrating Co-MIMO radio mesh arrays at about 35m increments. Due to the prevalence of multi-band 802.11ac/n WiFi COTS routers on the market that can run OpenWRT/LEDE distributions, this node type will likely be combined with the above node types.
-
~900MHz 802.11ah LoRaWAN nodes receiving wireless power and data from one of the above node types, or local micro-generation and battery storage, and coordinating Co-MIMO radio mesh arrays at about 60-300m increments. Due to the prevalence of low-cost SDR designs in the sub-GHz range, this node type will likely be combined with below node types. Gateways to >2GHz radio meshes will likely be combined with above node types.
-
~150MHz MURS nodes receiving wireless power and data from one of the above node types, or local micro-generation and battery storage, and orchestrating Co-MIMO radio mesh arrays at about 100m-1.5km increments. Due to the prevalence of low-cost SDR radio designs in the sub-GHz range, this node type will likely be combined with 900MHz and below node types. Gateways to >500MHz radio meshes will likely be combined to above node types. While at first glance MURS regulatory limitations don't lend themselves well to mesh data networking; low power MURS non-forwarding meshes still allow the higher bandwidth mesh types above to communicate geographic, timing, cryptographic key, PKI authentication, SMS-alike text block, and other meta data directly across mesh clusters that have no other direct data communication channels. This capability will greatly assist Web of Trust (WoT) metric discovery and inter-mesh-cluster authentication systems, help them avoid Man in the Middle (MitM) attack vectors, and can serve as a standardized Serval Mesh or similar text communications system during disaster recovery.
-
SDR capable nodes will make it possible to coordinate Co-MIMO arrays of radios using above ISM and wired node meshes for licensed and "light licensed" bands -- for example whitespaces, satellite K bands, UAV or balloon nodes, 3.6GHz security meshes, and licensed cellular pico stations or relays. SDR GPS and GLONASS receivers will also allow all nodes to more precisely coordinate unique node IP numbering and antenna phase timing without outside DHCP or NTP style interventions. Nodes lacking any direct GPS or GLONASS signals can be fed precise location and time data by 3 or more neighbor nodes via signal latency based triangulation. Noisy frequencies can also be pseudorandomly sampled by the SDR to generate cryptographically secure random numbers as needed.
-
Mobile device clients can be created that will simultaneously utilize all 6 unlicensed radio bands listed above, minimally in 2x2 MIMO configurations to traverse 2+ neighbor static nodes during travel without transition loss, optionally with low power 2x2 MIMO SDR antennae that can dynamically utilize other cellular licensed and "light licensed" bands in local mesh use. GPS and aGPS mobile devices can benefit from the same signal latency based triangulation methods for greater geolocation and UTC atomic time precesion.
Mesh network topologies with multipath routing lend themselves to secret sharing and redundancy methods that resist the old Man in the Middle (MitM) and Distributed Denial of Service (DDoS) attack vectors that make the Internet so insecure. Local or direct ownership of mesh infrastructure by residents where that infrastructure is installed, such as via community utility cooperatives, creates a natural resistance to powerful attackers, like malicious foreign or domestic State government actors and monopolist conglomerates.
Network Coding and Redundant Array of Inexpensive Links (RAIL) methods both provide data redundancy and splitting methods that avoid the need for repeated sends upon single-path failures, and resist MitM attacks along any one mesh path. They each have configurable M+N values, where M is the minimum number of data paths needed to recreate a full message stream, and N is the number of redundant paths that can be lost in transmission without requiring a resend. If any attacker has reliable control of nodes along less than M paths, it becomes impossible for them to reconstruct the original data stream even when its encryption key is already known to them. If the attacker has control of nodes along less than N paths, it becomes impossible for them to force a resend, and even less probable that any new resend paths will cross M or more of their controlled paths.
Local geographic IP numbering, which can both be enhanced and confirmed by neighbor latency triangulation based on physical values like the speed of light, is highly resistant to the same IP spoofing techniques that are all too easy on the ICANN centrally allocated ASN and IP number system. When these unique IP values are treated purely as relative position routing data, never as identification data, and Public Key Infrastructure (PKI) over Webs of Trust (WoT) are used for identification instead, the system as a whole becomes far more private and secure. In general, it is much easier to trust neighbors you can walk over to and interact with directly, rather than remote and uncaring institutions.
Measuring signal quality and latency over multiple radio bands also allows second-degree mesh node neighbors on smaller wavelengths to become first-degree (or direct) node neighbors on longer wavelengths. This attribute of light physics allows us to test all neighbors on all but the longest available wavelengths, to confirm that cryptographically secured challenge responses on direct connections match the responses bounced through closer neighbors on second-degree wavelength connections, third-degree wavelength connections, and so on. Optimistic tit-for-tat rival game methods in turn allow us to track these cryptographic relay challenge results over time, to be used as WoT metrics for future mesh path security evaluation. Cryptographically secure mesh cluster group-ledgers also allow neighbor node groups to coordinate inter-cluster trust metrics for automated WoT route maintenance, and provide an inherent incentive for trustworthy mesh peer transactions and forwarding.
Xerox PARC Named Data Networking (NDN) and NASA Delay Tolerant Networking (DTN) can be combined to replace both old centralized ICANN domain name allocation and private industry CDN infrastructure, in favor of decentralized keyword search and distributed version-controlled caching systems. Rather than forcing all data to "exist" on some central attackable file server, it instead "resides" on the network directly in the form of a universally distributed cache, identified by a cryptographically unique hash value and owner identifier (primary signer PKI ID). For the sake of consistency, all public-anonymous data can be given the same null PKI ID, with group pseudononymous data given private keys and owner ID's shared by the entire group. Whomever has access to the data can index its contents and metadata however they please, both individually and as shared access groups, so that finding it is a matter of a simple low-latency local mesh cluster metadata search. No complex URL or typo-ridden domain naming monopoly system is ever necessary! Retrieving the file or block of data is also a simple matter of contacting its owner (individual or group PKI ID and signer), learning where it probably has been cached via other recent nearby owner requests, and pulling it to a local cache from all remote cache sources where it is known to be available on open mesh paths. Future requests can be shortcut to the local mesh cluster cache, optionally asking the owner for any updates (at which time the hash ID also updates to the newest or HEAD version, and the prior version hash is kept as a parent or origin-branch identifier). If this data is not considered real-time latency dependent, such as an audio or video "call" stream, then owner requests and data location or packet responses can be transmitted via high-latency DTN systems on an opportunistic priority-tagged basis, such as mobile data-cache vehicles that travel between neighbor cities hourly, or high-orbit DTN satellites that pass by the sky daily. Smaller data requests and blocks, such as Serval Mesh text communications, will naturally be given higher priority than all larger data requests on finite DTN mobile caches.
Decentralizing both mesh connection hardware and data sources breeds security and reliability. Centralized authority data, including domain names and IP numbers, can always be lost or changed in transit, whether intentionally or not. Centralized tree or star topologies can be attacked at concentrated hubs to take down entire municipal regions. Decentralized data on mesh topologies can be authenticated from multiple paths and sources, leading to majority-vote or parliamentary style workarounds to direct attacks or byzantine faults. If we want to recreate a secure and private Internet from the ground up, we also need ownership and management to come from the user first, spread up and out among their direct neighbors, and eventually mesh the entire universe. In the process, you may discover your neighbor mesh cluster is even closer to Kevin Bacon's than you think.