My favourite football team was really at risk of relegation, then I created for me and my friends an MCMC bayesian simulator to estimate the relegation match by match.
It was an opportunity to get used in real life to some concepts (MCMC, Metropolis-Hastings, etc) that I always struggled to understand.
And my team got saved from relegation with an amounts of points very close to the number of points that my model forecasted
I also wrote an honeypot that emulate an Ollama instance: beside the attackers, it's funny how many people are looking for free inference. Somebody from Brazil try to use my honeypot to write to chapters of a book about traditional magic rituals.
My next step is to extract the data collected by this tool to extract IoC and malicious prompts and share them with the community.
In the same scope I wrote also an Ollama scanner: it fetch from Shodan the open Ollama instances, verify that they are reachable and check if they are real sending a dummy query.
Fiorentina, in Serie A.
After 11 matches we have only 4 points and we won the first match only in December: in the past nobody got saved with a so bad performances in the first matches.
I am an happy user of Meshstastic since more than one year (I have two active nodes and a third one is in the making). I am living in hilly countryside and the difficult that I have experienced is about reaching other nodes: with the standard antenna, I can barely connect with nodes in a 500 meters range, with a better antenna (coaxial collinear is the best IMHO) I can reach more than 10km.
I don't think the Meshstatic approach of "flooding" the network with all the messages can be scalable in the long run, they need to implement some sort of routing protocols (like BATMAN), but they are heavy and complex to implement
None of MeshCore, Meshtastic or Reticulum will scale well, especially not on top of a heavily constrained radio technology like LoRa. Flooding is inefficient for obvious reasons, AODV-esque routing (which MeshCore tries to do for DMs and Reticulum tries to do in general) is prone to almost-immediate path failure on unstable underlying transports, and the hidden node problem always bites on haphazard/unplanned mesh radio deployments where people show up with nodes in random locations on the same frequency.
The cracks are already extremely visible in MeshCore in the UK, where overheads from adverts and dropped packets from collisions mean it is already horrendously unreliable and most of the chatter in the Public channel is people sending test messages and being unsure whether anything they sent was ever heard by anyone.
Most other routing protocols (BATMAN included) are also not that well suited to situations where the underlying transport ends up asymmetric, e.g. one node can't hear others but it can be heard, and that's an extremely common occurrence/failure mode in wireless meshes like this. It's a difficult problem to solve with coordination between nodes, let alone without.
You have two conditions: not dense enough or too dense and the fractal nature of population mean you can have both. Radio bandwidth is a precious resource and if you think somehow that anything good will come out of sending a packet N times you are not likely to manage a wireless network successfully.
Italian here: as the first reading, I thought it was an usual decision against progress and technology.
But reading the article it seems a good sense rule: Lombardy was one of the most industrialized zones of Europe and now is migrating to a post industrial model. This law should force reusing the old and unused industrial spaces instead that wasting space in agricultural areas.
My master thesis was about an application of SSS to mesh networks: even if one of the node of the mesh was captured by an attacker and the secret retrieved from the node, it was impossible to crack the whole encryption.
When I was a poor student, Zaurus was probably my biggest tech gadget dream.
Unfortunately it was rare and expensive and I never had the opportunity to play with it.
It was a hype in the Linux community at the time. Around that time I moved from study to job and had a bit of cash to spare. Couldn't resist a good deal for a second hand SL-5500, but I think it was ahead of its time and the potential not fully realized.
It was also not really open source. You could compile your own distro, but nothing ever produced by the Zaurus community would ever offer as much functionality as the proprietary Linux desktop from Sharp on the 5500 at least.
6 miles seems a very optimistic estimation: 2.4Ghz propagation is very reduced by obstacles like buildings or trees and at that frequency the atmospheric water (fog, rain, humidity) have a big impact on propagation. And you need also to consider that 2.4Ghz is a very polluted band, then the noise floor is significatevly higher than in the 865/915 Mhz.
Moreover at 2.4Ghz the Fresnel window is smaller and the risk of multipath fading is higher.
LORA uses a sub noise-floor link budget. It allows some pretty crazy performance, at the expense of massive speed losses. Like 203kbps for LoRa vs 1,376,000kbps for WiFi lol.(max phy speeds, ymmmv).
WiFi sensitivity is about -90dB, while LoRa sensitivity is around-150dB…. So that’s about a million times more sensitive. So you need about a million times more signal strength to use low bandwidth WiFi (still impossibly fast by LoRa standards) than to use low bandwidth LoRa.
Those are radio specifications. Real links require about 10db more to get any kind of reliability, but the comparison stands.
> 2.4Ghz propagation is very reduced by obstacles like buildings
I never did much 2.4ghz stuff because that was what rich people did, or people mad enough to modify microwave oven magnetrons. However I was always under the impression that freespace loss on 2.4 was terrible. but it turns out its "only" ~9db more than 865
Worth mentioning that 2.4GHz has a lot more attenuation due to clutter than 900MHz. Your problem is usually buildings and non line of site transmission paths. When the signal has to pass through and bounce off things your link budget takes a big hit.
There is the idea of the path loss exponent. In a vacuum it's 2.0. 900Mhz with clutter it's -2.5 At 2.4 MHz it's -3 and -5.8 it's -3.5.
Other downside for higher spreading factor spectrum is data rate drops which results in longer packets. Longer packets means more energy per packet and a higher chance someone else will blow your packet out of the water.
You've been able to buy 900 and 2.4GHz transceivers for the last 20 years.
Not much. While this is technically LoRa on 2.4GHz (which is not new), most people will associate LoRa with significantly longer range and LoRa 2.4 can do.
Also for a student: I was at university 20 years ago and there was no possibility to "hack" RF devices.
Right now, with less than 200€ (a NanoVNA and a good SDR), you can do almost everything.
Combined with a good directional coupler and GPSDO, people do build very high-end pieces of equipment. Not user friendly for most students, but removes a lot of traditional DSP complexity.
Current design limits on bit-resolution and bandwidth are usually just people avoiding getting hit with export restrictions.
The Pluto SDR rich software ecosystem is likely a better option for students:
In Italy we have an "indipendent research lab" that become really famous for a study that demonstrates that aspartame may cause cancer.
The same institute published few years later a study about 5G emissions that may cause cancer.
“ However, a number of major issues with the study were identified by the Panel which made interpretation of the findings difficult. Notably, a high background incidence of chronic inflammatory disease in the lung and other organs was observed in all the animal groups including controls which did not receive aspartame, as reported by the European Ramazzini Foundation. This was considered to be a major confounding factor.”
Not a medical professional, but inflammation is something different from cancer that they mentioned in their website.
And we need to understand also the trial scenario: in the one about 5G they expose rats for more than 20 hours to a radio power more than 10 the law limits.
I think you and I agree. This is about the Italian cancer aspartame study you referenced (Ctrl-f on cancer). This is EFSA saying the study has major issues and reiterating that aspartame is safe.
I'm going to start a research lab that releases dubious conclusions so I can then be hired by other companies to "bad mouth" their product just so people on the internet can link the two and conclude that their product is actually good.
I know this lab! Ramazi Institute or something right?
We covered it in this podcast I used to produce (not explicitly the subject, they came up re: artificial sweetener studies and we explored them a bit). They’re very good at appearing legitimate while pushing wild claims.
There is a waste area in the Italian society that is very prone to the conspiracy theory. Some famous journalists and some TV shows are very good in spreading this news.
In the past, a party (M5S, now pivoting to a left wing, populistic, pro Putin movement) took 34% at the election, just taking advantage of those crazy ideas.
I also wrote an honeypot that emulate an Ollama instance: beside the attackers, it's funny how many people are looking for free inference. Somebody from Brazil try to use my honeypot to write to chapters of a book about traditional magic rituals. My next step is to extract the data collected by this tool to extract IoC and malicious prompts and share them with the community.
In the same scope I wrote also an Ollama scanner: it fetch from Shodan the open Ollama instances, verify that they are reachable and check if they are real sending a dummy query.
reply